Digital Health

AI Diet Recommendations for Adolescents Show Significant Nutritional Gaps, Study Finds

Key Takeaways:

• AI-generated diet plans consistently underestimated energy and key macronutrients required by adolescents
  
• Macronutrient balance was frequently misaligned with clinical guidelines, with lower carbohydrates and higher fat and protein levels
  
• Researchers caution that AI tools should not replace dietitians for adolescent nutrition without professional oversight
  
  
  

Growing demand for accessible nutrition support

Artificial intelligence is increasingly being used to support dietary planning, particularly in areas where access to qualified professionals is limited. However, a new study published in Frontiers in Nutrition raises important concerns about the reliability of these tools when applied to adolescents living with overweight or obesity.

Globally, adolescent overweight and obesity are rising at pace, affecting an estimated 390 million young people in 2022. In many regions, this now represents the most common form of malnutrition. Excess body weight in adolescence is associated with a range of adverse health outcomes, including type 2 diabetes, dyslipidaemia, hypertension, and sleep apnoea. It also increases the likelihood of obesity in adulthood and is linked to reduced quality of life.

Alongside physical health risks, adolescents may experience body image concerns and engage in harmful weight control behaviours such as self-induced vomiting or misuse of laxatives.

Dietary modification remains central to improving outcomes. Dietitians play a key role in delivering tailored, evidence-based nutrition plans aligned with established guidelines. However, limited access and workforce pressures can restrict the availability of personalised support.

AI tools, including chatbots and large language models, are increasingly being explored as a way to bridge this gap. While they can provide general dietary guidance, concerns remain about their accuracy, safety, and ability to replicate the individualised care provided by trained professionals.

Study design – comparing AI models with dietitian plans

To better understand the role of AI in adolescent nutrition, researchers conducted a direct comparison between AI-generated diet plans and those created by a dietitian.

Five AI systems were evaluated: ChatGPT-4o, Gemini 2.5 Pro, Claude 4.1, Bing Chat-5GPT, and Perplexity. Across two sessions, these models generated a total of 60 diet plans. Each plan covered three days and was based on four standardised adolescent profiles, including boys and girls living with overweight or obesity.

These AI-generated plans were compared with dietitian-designed one-day plans developed in line with established nutritional recommendations. The reference plans followed a macronutrient distribution of:

• 45–50 % carbohydrates
• 30–35 % fat
• 15–20 % protein

The researchers then analysed energy intake, macronutrient composition, micronutrient content, safety, and feasibility.

Consistent underestimation of energy and macronutrients

The findings revealed a clear and consistent pattern across all AI models. Diet plans generated by AI underestimated both total energy intake and key macronutrients when compared with dietitian-designed plans.

On average:

• Energy intake was lower by 695 kcal
• Protein intake was reduced by 20 g
• Fat intake was reduced by 16 g
• Carbohydrate intake was reduced by 115 g

Given the high energy demands of adolescence, such deficits could have meaningful clinical implications, particularly for growth, development, and overall health.

Macronutrient imbalance – a shift away from guidelines

Beyond total intake, the balance of macronutrients was also significantly altered in AI-generated plans.

Some AI models recommended:

• Protein intake up to 23.7 %
• Fat intake up to 44.5 %

Both values exceeded recommended levels. In contrast, carbohydrate intake accounted for no more than 36.3 %, falling below guideline recommendations.

Dietitian-designed plans, by comparison, remained closely aligned with clinical standards:

• Carbohydrates: 44 %–46 %
• Protein: 18 %–20 %
• Fat: 36 %–37 %

The authors noted:

“This pattern illustrates a systematic shift across all AI models to lower CHO, higher protein, and higher lipid meal structures, indicating that the macronutrient balance, not just the amount of gram-based nutrients, is significantly disrupted in AI-generated plans.”

Researchers suggest that AI models may be influenced by popular dietary trends, such as low-carbohydrate or ketogenic approaches, rather than evidence-based adolescent nutrition guidelines. This shift may pose risks during a critical period of physical and cognitive development.

Micronutrient variability raises additional concerns

In addition to macronutrient discrepancies, the study identified significant variability in micronutrient composition across AI-generated plans.

No model consistently matched the dietitian-designed reference diet across all nutrients. This inconsistency raises concerns about potential micronutrient deficiencies, which could further compromise adolescent health.

The findings suggest that AI tools currently lack the technical precision required to accurately estimate both macro- and micronutrient needs in personalised dietary plans for adolescents.

Strengths and limitations of the study

The study offers several notable strengths. It evaluated multiple AI models, allowing for robust comparison across systems. The use of three-day diet plans enabled identification of consistent patterns rather than isolated outputs. Dietitian-designed plans provided a credible clinical benchmark, and the inclusion of both macro- and micronutrient analysis allowed for a comprehensive assessment of dietary quality.

However, there are limitations to consider. The findings are specific to the models tested, which are rapidly evolving. Standardised adolescent profiles may not fully capture real-world complexity, limiting personalisation. The use of simulated scenarios rather than real-life behaviours may reduce ecological validity. Additionally, prompts were standardised and delivered in a single language, which may limit generalisability across populations.

Implications for clinical practice and AI use

The study highlights important risks associated with the unsupervised use of AI for adolescent dietary planning.

As the authors conclude:
“AI models have exhibited clinically significant deviations in diet plans for adolescents at both macro and micro levels.”

These deviations include consistently lower energy and carbohydrate recommendations compared with dietitian-designed plans.

Until these limitations are addressed, AI-generated diet plans should be used with caution. They may serve as a supplementary tool under professional supervision, but they are not currently a safe or reliable substitute for qualified dietary guidance in adolescents.

Read More

Virtual Care Could Reduce Hospital Admissions in Severe Eating Disorders, Study Finds

Key Takeaways:

• A fully virtual, multidisciplinary treatment programme demonstrated strong engagement and positive clinical outcomes for adults living with severe eating disorders
  
• Structured online support during high-risk transition periods may help reduce hospital admissions and sustain recovery following discharge
  
• The model highlights the potential for digitally delivered, evidence-based care to bridge gaps between inpatient and community services
  
  
  

Evaluating a new approach to severe eating disorder care

An evaluation conducted by Oxford Health NHS Foundation Trust has examined whether intensive, fully virtual treatment can effectively support people living with severe eating disorders. The study, published in the Journal of Eating Disorders, focused on a service known as Step Care, developed through the HOPE Provider Collaborative.

Researchers describe this as the first prospective study to investigate a completely virtual, intensive treatment model using multidisciplinary enhanced cognitive behavioural therapy (CBT-E). The programme is designed to support individuals as they transition between inpatient treatment and community-based care.

This period of transition is widely recognised as a critical phase in recovery. People living with severe eating disorders face a particularly high risk of relapse shortly after discharge from hospital, especially within the first two months. The study therefore explored whether structured virtual support during this vulnerable period could help maintain recovery and reduce the likelihood of readmission.

Addressing gaps in existing services

Step Care was developed in response to well-documented challenges within eating disorder services. These include fragmented transitions between inpatient and community care, repeated hospital admissions, and limited access to intensive day treatment.

The service is delivered entirely online and brings together a multidisciplinary team, including professionals from psychology, nursing, dietetics, and art therapy. This integrated approach aims to provide consistent and coordinated care across different stages of recovery.

Step Care operates through three distinct pathways:

• Starting Well – for individuals at risk of requiring hospital admission, with a focus on prevention
  
• Staying Well – for those recently discharged from inpatient care, supporting early recovery
  
• Working towards Recovery – for individuals who have begun restoring weight and are focusing on longer-term recovery
  

Lucy Gardner, professional lead dietitian within the Step Care service, highlighted the importance of integrating nutritional support within a broader therapeutic framework:

“Nutrition plays a crucial role in mental health, yet access to the right level of dietetic support is often inconsistent,” she said. “Our model offers a clear, evidence-informed way to tailor dietetic input to individual need, delivering CBT-E virtually as part of a multidisciplinary team.”

Positive outcomes across key measures

The evaluation reported high levels of engagement and programme completion, including among individuals who had been living with eating disorders for an extended period.

Participants within the Starting Well pathway experienced significant improvements across several clinical and psychological measures, including:

• Body mass index (BMI)
  
• Eating disorder symptoms
  
• Psychosocial impairment
  
• Mood
  

Importantly, most individuals in this group were able to avoid hospital admission during the course of the programme.

For those in the Staying Well pathway, outcomes were also encouraging. Participants maintained their weight and experienced a reduction in the overall impact of their illness during a period typically associated with high relapse risk. Unplanned hospital admissions were reported to be rare, and many individuals were successfully supported in transitioning to community-based care.

Sharon Ryan, nurse lead within the Step Care service, emphasised the importance of this post-discharge phase:

“The weeks after leaving hospital are often the most fragile,” she said. “Step Care provides consistent multi-disciplinary support at that point, helping people maintain their recovery with support to feel safe and confident out of hospital.”

Implications for future care models

The findings suggest that intensive, evidence-based treatment for severe eating disorders can be delivered effectively in a virtual format. This approach may offer a valuable additional option for supporting individuals at home, particularly during critical transition periods.

Agnes Ayton, clinical lead for the HOPE Provider Collaborative, explained the underlying aim of the service:

“Step Care was designed to bridge the gap between inpatient and community services,” she said. “The findings show that intensive, evidence-based treatment can be delivered safely online, providing continuity of care at a time when people are most vulnerable.”

She also noted that both engagement and clinical outcomes were encouraging, including among individuals who had experienced long-term illness.

A complement to existing services

The authors conclude that virtual programmes such as Step Care may serve as an important complement to traditional inpatient and community services. By providing structured, multidisciplinary support during high-risk periods, these models have the potential to enhance continuity of care and support sustained recovery for people living with severe eating disorders.

As healthcare systems continue to explore digital and hybrid models of care, this study adds to a growing body of evidence suggesting that virtual interventions can play a meaningful role in complex, long-term conditions.

Read More

AI in Healthcare: Promise, Pitfalls, and the Risk of Misguided Medical Advice

Key Takeaways:

• People using AI for health advice often struggle to interpret and communicate symptoms effectively, leading to incorrect conclusions in many cases.
  
• Even when AI identifies a condition correctly, it may fail to recommend appropriate urgency, particularly in time-sensitive or complex scenarios.
  
• Clinicians see value in AI as a supportive tool, but stress that it should complement, not replace, professional medical care.
  
  
  

AI becomes a common source of health information

As technology companies continue to develop platforms tailored for healthcare consultation, artificial intelligence is becoming an increasingly influential part of how people make decisions about their health. According to OpenAI, more than 40 million people use ChatGPT each day to seek health-related information.

However, emerging research suggests that while these tools offer unprecedented access to medical knowledge, they may also mislead users in certain contexts.

Challenges in how people use AI for medical queries

One of the central issues identified by researchers is not only the capability of AI systems, but how individuals interact with them. Many people lack the knowledge required to communicate symptoms accurately or comprehensively.

A recent study published in Nature Medicine attempted to replicate real-world use of AI chatbots. Participants were given medical scenarios and asked to consult AI tools. The results highlighted notable limitations:

• Participants correctly identified the condition only about one-third of the time.
  
• Just 43% made the correct decision regarding next steps, such as whether to seek emergency care or remain at home.
  

“People don’t know what they are supposed to be telling the model,” says Andrew Bean, who studies AI systems at Oxford University and was one of the authors on this study.

Bean explains that effective use of AI often depends on precise wording. “Doctors are trained to ask you questions about symptoms you might not have realised you should have mentioned,” says Bean.

Small differences in input can lead to dangerous outcomes

The study demonstrated how subtle differences in language can significantly alter the advice provided by AI systems.

In one example, two individuals described the same clinical scenario slightly differently. One described experiencing “the worst headache I’ve ever had” and was advised to go to the emergency room immediately. The other, who did not include that specific phrasing, was advised to take aspirin and remain at home.

“Turns out this was actually a life-threatening condition,” says Bean.

This highlights a critical limitation: AI systems rely heavily on the information they are given, and may not prompt for missing but clinically important details in the way a trained clinician would.

When AI gets the diagnosis right but the advice wrong

Even when AI tools successfully identify a medical condition, they may still provide inappropriate guidance regarding urgency or next steps.

In a separate study, researchers evaluated how AI systems responded to a range of medical scenarios. They found that in 52% of emergency cases, the tools “under-triaged” – treating conditions as less serious than they actually were.

In one case, the AI failed to direct a hypothetical patient experiencing diabetic ketoacidosis and impending respiratory failure – both life-threatening conditions – to seek emergency care.

“When there was a textbook medical emergency, ChatGPT got it right,” said Girish Nadkarni, a doctor and AI researcher at Mount Sinai who is an author on the study. However, he noted that performance declined in more complex situations, particularly where timing was critical. In such cases, the system often misjudged how urgently care was required.

An OpenAI spokesperson responded by stating that the study did not reflect typical real-world usage and that it evaluated an older version of ChatGPT, which the company says has since been improved to address some of these concerns.

The role of AI in supporting patient understanding

Despite these concerns, many clinicians believe that AI tools can still play a constructive role in healthcare, particularly in improving patient understanding and engagement.

“I encourage patients to use these tools,” says Robert Wachter, a doctor at UC San Francisco and author of the recently published book, A Giant Leap: How AI Is Transforming Health Care and What That Means for Our Future.

Wachter points out that barriers to accessing healthcare – including cost and availability – mean that AI can sometimes provide a useful alternative source of information. “The advice you get from the tools is substantially better than nothing and better than what you would get from your second cousin,” says Wachter.

However, he emphasises that AI should never be viewed as a substitute for professional medical care.

Enhancing, not replacing, the doctor–patient relationship

Experts suggest that AI is most valuable when used alongside traditional healthcare, rather than in place of it.

Adam Rodman, a hospitalist and researcher at Harvard Medical School, advises against using AI tools to assess emergency situations. Instead, he sees their greatest benefit in preparing for or reflecting on medical consultations.

“A good time to use a large language model is when you’re about to go see a doctor – or after you see your doctor,” says Rodman.

He explains that AI can help people better understand their condition, ask more informed questions, and make more effective use of time during appointments. This can support a more collaborative relationship between patients and clinicians.

“There are no downsides to better understanding your health,” says Rodman.

The future of AI in healthcare

Healthcare professionals broadly agree that AI is now firmly embedded within modern medicine and will continue to evolve alongside clinical practice.

“ My hope is that you might see AI as an extension of a human relationship,” says Rodman. He envisions a future in which both clinicians and patients work with AI to improve communication and navigate healthcare systems more efficiently.

However, he also raises concerns about potential unintended consequences. One particular risk is the possibility that people may receive serious or distressing diagnoses – such as cancer – directly from an AI system, rather than from a clinician.

Research suggests that when healthcare becomes more transactional or resembles a marketplace, trust in clinicians may decline.

 ”What I hope is that this technology can be used in a way that enhances humanity in medicine,” says Rodman “and not in a way that cuts out the doctor-patient relationship.”

Conclusion

Artificial intelligence is rapidly transforming access to health information, offering both opportunities and risks. While these tools can enhance understanding and support more informed decision-making, their limitations – particularly in how they interpret incomplete or imprecise input – mean they must be used with caution.

Ultimately, AI has the potential to strengthen healthcare delivery, but only if it is integrated in a way that supports, rather than replaces, the human relationships at the heart of medicine.

Read More

First UK Long-Distance Robotic Surgery Connects London Surgeon with Gibraltar Patient

 Key Takeaways:

• A London-based surgeon has performed the UK’s first long-distance robotic cancer surgery on a patient in Gibraltar, marking a major milestone in telesurgery
  
• The procedure demonstrated minimal delay and high precision, suggesting remote surgery could expand access to specialist care in underserved regions
  
• Patients living far from specialist centres may benefit from reduced travel, lower costs and improved continuity of care
  
  
  

A landmark moment in remote surgery

A surgeon based in London has carried out what is believed to be the United Kingdom’s first long-distance robotic surgical procedure, operating on a patient located approximately 1,500 miles (2,400 km) away in Gibraltar.

Professor Prokar Dasgupta, a leading robotic urological surgeon, performed a prostate removal on 62-year-old Paul Buxton. Reflecting on the experience, he said it felt “almost as if I was there”, despite the geographical distance between surgeon and patient.

For Buxton, who is living with prostate cancer, the decision to participate in the procedure was straightforward. He described it as a “no-brainer” and an opportunity to become “part of medical history”.

Expanding access to specialist care

The development of long-distance robotic surgery is seen as a potential solution to longstanding challenges in healthcare access, particularly for people living in remote or underserved regions.

Such approaches could reduce the “vast expense and inconvenience” associated with travelling for specialist treatment, while enabling patients to receive care closer to home.

This milestone builds on previous advances involving UK-based surgical teams. Earlier work included a transatlantic robotic stroke procedure conducted over a distance of 4,000 miles on a cadaver – a body donated to science – which demonstrated that long-distance surgery was technically feasible.

A patient’s perspective

Buxton, originally from Burnham-on-Sea in Somerset, has lived in Gibraltar for four decades. As a British Overseas Territory, Gibraltar has limited healthcare infrastructure, with only one hospital – St Bernard’s Hospital at Europort. Patients requiring more complex care often need to travel abroad, commonly to the United Kingdom for NHS treatment if eligible.

Following his prostate cancer diagnosis shortly after Christmas, Buxton initially expected to join an NHS waiting list. However, he chose instead to take part in the remote surgery trial.

“A lot of people actually said to me: ‘You’re not going to do it, are you?’”

“I thought, I’m giving something back here,” he said.

Buxton also highlighted the practical advantages of the approach:

“If I hadn’t gone for the telesurgery in Gibraltar, then I would have had to have flown to London, I would have had to go on the NHS waiting list, get the procedure done and I would have probably been in London for three weeks.

“So I thought: ‘This is a no-brainer’.

“It is pioneering for Gibraltar, because you don’t need to leave Gibraltar.”

Following the operation on 11 February, he reported a positive recovery, stating he was “really well looked after” and “feeling fantastic”.

How the technology works

The procedure was conducted from The London Clinic using a robotic surgical system equipped with a high-definition 3D camera and four robotic arms. These were controlled remotely via a surgical console.

The connection between London and Gibraltar was enabled through fibre-optic cables, supported by a backup 5G link. The system achieved an extremely low latency, with a delay of just 0.06 seconds, allowing for precise and responsive control.

A surgical team in Gibraltar remained on standby throughout the operation to intervene if necessary, although the connection remained stable for the duration of the procedure.

The operation utilised the Toumai Robotic System and was delivered through a collaboration between The London Clinic and the Gibraltar Health Authority.

Looking ahead: scaling telesurgery

Professor Dasgupta emphasised the broader implications of the innovation:

“This gives us the opportunity to treat patients in remote areas and smaller communities by literally being able to take the best surgeon anywhere.”

The procedure forms part of an initial series of test cases. A second operation involving a 52-year-old patient in Gibraltar was carried out on 4 March, with a further procedure scheduled for 14 March.

The upcoming operation will be live-streamed to 20,000 leading urological surgeons attending the European Association of Urology congress, highlighting the global interest in this emerging field.

Reflecting on the future, Dasgupta added:

“I think it is very, very exciting, the humanitarian benefit is going to be significant.”

Alignment with broader surgical trends

This development sits alongside wider efforts to expand the use of robotic-assisted surgery within the NHS. Current ambitions include scaling up to 500,000 robot-supported operations annually by 2035.

While the NHS is prioritising local access to robotic surgery, advances in telesurgery suggest a complementary pathway – one that could extend specialist expertise beyond physical borders and reshape how surgical care is delivered globally.

Read More

‘Shadow AI’ on the Rise in Healthcare as Clinicians Turn to Unauthorised Tools to Improve Workflows

Key Takeaways: 

• A survey of healthcare professionals found that 57% have encountered or used unauthorised artificial intelligence tools in their organisations, highlighting the growing presence of so-called “shadow AI” in healthcare settings.
  
• Many clinicians and administrators report using these tools to improve efficiency, analyse data, and manage administrative tasks, particularly when approved solutions or clear guidance are lacking.
  
• While most respondents believe AI will significantly improve healthcare within five years, concerns about patient safety, data privacy, and security risks remain widespread.
  
  
  

Unauthorised AI tools emerging in healthcare workplaces

A new survey suggests that artificial intelligence tools are already being used in healthcare organisations in ways that fall outside formal governance structures. According to the findings, a significant proportion of healthcare professionals have either encountered or used AI tools that have not been authorised by their employer.

The survey, conducted by Wolters Kluwer Health, gathered responses from 518 healthcare professionals, including both clinical providers and administrators. The research was carried out in December 2025 and was released publicly last week.

Overall, the findings indicate that four in ten healthcare professionals reported encountering unauthorised AI tools within their organisation, while 17% acknowledged personally using such tools.

When responses were analysed by professional role, 15% of physicians admitted to using an unauthorised AI tool, compared with 19% of administrators. In addition, one in ten respondents reported using an unauthorised AI tool in connection with direct patient care.

The report refers to the unauthorised adoption of artificial intelligence tools in professional environments as “shadow AI.”

Why healthcare staff turn to unauthorised AI

The survey findings suggest that healthcare professionals are often motivated by practical needs rather than deliberate attempts to bypass organisational policies.

According to the report:

“Clinical and administrative teams want to adhere to rules surrounding AI usage, but if the organization hasn’t provided guidance or approved solutions, they’ll experiment with generic tools to improve their workflows.”

Many respondents indicated that the absence of formal guidance or approved AI platforms has encouraged individuals to explore publicly available tools on their own.

The most frequently cited motivation for using unauthorised AI tools was the need to accelerate workflows and improve efficiency. Approximately half of respondents identified faster workflows as the primary reason for using these tools.

However, the survey also revealed differences in how clinical and administrative staff tend to use AI technologies.

Administrators were more likely to employ AI tools for operational or analytical tasks such as:

• Data analysis
  
• Predictive analytics
  
• Administrative processes
  

Healthcare providers, meanwhile, reported using AI for activities such as:

• Data analysis
  
• Patient scheduling
  
• Patient engagement tasks
  

The findings also indicate that clinicians were more likely than administrators to experiment with AI tools out of curiosity.

Governance and policy development remain uneven

The survey results highlight a notable imbalance in how different professional groups participate in the development of AI policies within healthcare organisations.

According to the report, administrators were three times more likely than clinical providers to be actively involved in developing AI governance policies.

Specifically:

• 30% of administrators reported involvement in AI policy development
  
• Only 9% of providers said they had participated in such efforts
  

This difference suggests that policy ownership around AI adoption may currently be concentrated within administrative leadership rather than clinical teams.

Administrators also reported greater familiarity with their organisation’s AI policies compared with providers, although awareness varied across both groups.

Security and privacy risks associated with “shadow AI”

The use of unauthorised AI tools raises important concerns about data security, privacy protection, and governance oversight.

The Wolters Kluwer report notes that inconsistent or unsanctioned AI usage can expose organisations to potential vulnerabilities. Without clear oversight, the integration of external AI tools may lead to data privacy violations, security breaches, or inappropriate handling of sensitive information.

To illustrate these risks, the report references a 2025 study by IBM, which found that 97% of organisations that experienced an AI-related security incident lacked adequate AI access controls.

Security incidents involving AI systems can have significant consequences, including financial losses, operational disruption, and damage to public trust.

Healthcare professionals remain optimistic about AI’s future

Despite concerns about governance and security, the survey indicates that most healthcare professionals remain broadly optimistic about the long-term role of artificial intelligence in healthcare.

Nearly 90% of respondents said they believe AI will significantly improve healthcare within the next five years. Administrators were found to be slightly more optimistic than clinical providers about the potential benefits of the technology.

At the same time, respondents recognised that AI implementation carries important risks that must be addressed.

Patient safety was identified by around half of respondents as the most significant risk associated with AI adoption.

Meanwhile, nearly half of respondents also expressed concerns about data privacy risks.

These findings suggest that healthcare professionals recognise both the transformative potential of artificial intelligence and the need for careful governance, clear guidance, and secure systems.

Addressing the rise of “shadow AI”

The report concludes that addressing the growth of shadow AI requires organisations to understand why staff are turning to unauthorised tools rather than focusing solely on restricting access.

According to the report:

“Ultimately, addressing shadow AI is not about restricting access to productivity tools. Leaders must understand why teams are using unsanctioned tools and which challenges they’re trying to solve, and then identify enterprise-level tools that can accomplish these goals safely and securely.”

As artificial intelligence becomes increasingly embedded in healthcare workflows, organisations may need to develop clearer policies, provide approved tools, and involve both clinical and administrative staff in governance decisions.

Such measures may help ensure that the benefits of AI can be realised while protecting patient safety, safeguarding sensitive data, and maintaining organisational trust.

Read More

New Self-Aware Biosensor System Could Improve Reliability of Wearable Medical Devices

Key Takeaways:

• Researchers have developed a new biosensor monitoring system that can rapidly detect when electrodes in wearable medical devices begin to detach from the skin.
  
• The technology evaluates the quality of digital signals transmitted between electrodes through the body, allowing direct monitoring of electrode contact.
  
• Early testing shows the system can identify early signs of electrode failure that conventional monitoring approaches often miss, potentially improving the reliability of digital health monitoring.
  
  
  

Advances in wearable biosensing for modern healthcare

Smart biomedical technologies are increasingly shaping the future of healthcare. A growing number of digital health tools rely on skin-mounted biosensors that collect detailed physiological data directly from the human body. These devices are commonly used in applications such as heart rhythm monitoring, remote patient monitoring, and long-term health tracking.

As these technologies become more widely adopted in clinical practice and home-based healthcare settings, the accuracy and reliability of the signals they collect become critically important. If the sensors or electrodes attached to the skin begin to loosen or detach, the data captured by the device can become unreliable.

To address this challenge, a research team at King Abdullah University of Science and Technology (KAUST) has developed a new system designed to detect electrode detachment in real time. The technology enables medical devices to identify when electrodes begin to lose proper contact with the skin, allowing clinicians and users to maintain accurate physiological monitoring.

The study describing the system was published in the journal Results in Engineering.

Limitations of traditional electrode monitoring methods

Many wearable medical devices rely on electrodes placed on the skin to detect electrical signals produced by the body, such as those generated by the heart. However, ensuring that these electrodes remain properly attached throughout monitoring can be difficult.

Conventional systems typically rely on indirect methods to determine electrode integrity, such as measuring electrical impedance or using other proxy indicators. These techniques were developed many years ago and often assume stable monitoring conditions.

According to the researchers, these assumptions do not always reflect real-world use.

“Traditional methods for checking whether medical electrodes are properly attached, based on impedance or indirect monitoring, were developed many years ago and assume relatively stable conditions,” explains Rajat Kumar, a student working in the laboratory of Ahmed Eltawil, who led the research.

In everyday situations, however, people move, perspire, and change position. These normal activities can cause electrodes to loosen slightly or temporarily lose contact with the skin.

Such intermittent disruptions can be difficult for conventional monitoring approaches to detect.

“This is especially problematic for home-based wearable medical devices, where poor electrode contact may go unnoticed for long periods, leading to inaccurate data being recorded and relied upon,” says Abdelhay Ali, a postdoctoral researcher in Eltawil’s research group.

Rethinking the body as part of the monitoring system

To overcome these limitations, the KAUST team reconsidered how electrodes interact with the body during monitoring.

Instead of viewing the human body purely as a source of interference in electrical measurements, the researchers explored whether it could become part of the detection mechanism itself.

Eltawil describes this shift in perspective:
“Instead of treating the body as something that interferes with measurements, we considered whether it could be part of the solution.”

Previous research has shown that very small electrical signals can safely travel through the body. The researchers realised that this property could be used to evaluate the condition of electrode attachments.

“We realized that if electrodes could exchange digital signals through the body, then the quality of that communication would directly reflect how well the electrodes were attached,” Kumar says.

If the electrodes remain firmly attached, the signals between them would be transmitted clearly. If the electrodes begin to loosen, the signal quality would deteriorate.

How the self-aware monitoring system operates

To test this concept, the team developed a monitoring system built around a custom-designed microchip created at KAUST.

The system works by sending very small digital signals between electrodes positioned at different locations on the body. These signals pass through the body and are then received by other electrodes.

A small processing unit analyses how well the signals are received.

According to Ali, the signal quality provides a direct indication of electrode contact:

“Clear signals indicate good electrode skin contact; small errors indicate weakening contact; and missing signals indicate disconnection.”

In addition to the chip and signal-processing unit, the system includes a control component that manages the electrode-checking sequence. This allows the device to automatically evaluate multiple electrodes in sequence without interrupting the primary medical measurements being performed by the device.

Testing the system on human skin

To evaluate the system’s effectiveness, the researchers conducted experiments using electrodes placed on human skin.

The testing showed that the system could reliably distinguish between several different conditions of electrode attachment, including:

• Firmly attached electrodes
  
• Partially loosened electrodes
  
• Electrodes that intermittently lose contact with the skin
  
• Completely disconnected electrodes
  

Importantly, the system demonstrated the ability to detect early stages of contact degradation before full disconnection occurs.

“Importantly, the system detected the early signs of contact degradation that traditional methods often miss,” Kumar says.

This early detection could be particularly valuable in wearable health monitoring devices that operate continuously over long periods.

Potential benefits for long-term wearable monitoring

A key feature of the new system is its very low power consumption, which makes it suitable for wearable technologies that must operate continuously for hours or days at a time.

Ali explains that this efficiency could make the technology practical for real-world use.

“The system’s very low power consumption should enable practical integration with wearable medical devices that need to run continuously for long periods,” he says.

He also notes that the design could be incorporated into existing devices with minimal modifications.

“These components form a compact and efficient solution that can be added to existing medical devices with minimal changes.”

Towards fully integrated wearable medical devices

Although the current system has been demonstrated in laboratory testing, the research team is now working to advance the technology further.

Their next goal is to develop a fully integrated single-chip system capable of monitoring many electrodes simultaneously.

Such a system could be used in a range of clinical monitoring devices, including multi-lead electrocardiogram (ECG) monitors and other wearable biosensing platforms used in both hospital and home environments.

Eltawil emphasises the broader aim of translating the technology into practical healthcare solutions.

“Ultimately, our goal is to translate this KAUST-developed technology into practical medical devices that are more reliable, more trustworthy, and better suited for continuous health monitoring in the clinic and at home,” he says.

Read More

Apple Watch Use in Older Adults Linked to Fourfold Increase in Atrial Fibrillation Detection, Study Finds

Key Takeaways:

• Older adults using an Apple Watch were four times more likely to be diagnosed with atrial fibrillation than those receiving standard care.
  
• Over half of the people diagnosed in the smartwatch group had no outward symptoms and were identified through watch alerts.
  
• Researchers suggest that smartwatch screening could reduce stroke risk and healthcare costs by accelerating diagnosis.
  
  
  

Smartwatches and fitness trackers have increasingly incorporated electrocardiogram functionality, allowing users to record heart rhythm data from their wrists. Among these devices, the Apple Watch has emerged as one of the most advanced consumer wearables in this space, capable of detecting irregular heart rhythms and, in some models, identifying indicators associated with raised blood pressure.

While these features have received certification from the US Food and Drug Administration for consumer use, the Apple Watch is not formally approved as a medical diagnostic device for clinical settings. Its purpose is to alert individuals to possible irregularities, prompting them to seek professional medical assessment for confirmation and formal diagnosis. In some cases, such alerts have led to early intervention, including reports of potentially life-saving outcomes.

New research now strengthens the case for wearable technology as a supportive diagnostic tool, particularly in detecting atrial fibrillation – a common heart rhythm disorder that can significantly increase the risk of stroke if left untreated.

Study design and participant profile

The research was conducted by investigators at Amsterdam University Medical Center. The team enrolled 437 adults aged over 65 who were considered to be at elevated risk of stroke.

Participants were divided into two groups:

• 219 individuals were provided with an Apple Watch for heart rhythm monitoring
  
• 218 individuals received standard care without smartwatch monitoring
  

All participants were assessed for atrial fibrillation, a condition that can be intermittent and frequently asymptomatic, making it challenging to detect through routine clinical encounters alone.

Fourfold increase in diagnoses

Among those using the Apple Watch, 21 individuals were diagnosed with atrial fibrillation and subsequently received medical care. Notably, 57 per cent of those diagnosed in the smartwatch group had no outward symptoms beyond what was indicated on their device.

In contrast, only five individuals in the standard care group were diagnosed with atrial fibrillation. All of these individuals were symptomatic at the time of diagnosis.

Overall, smartwatch monitoring identified four times as many people who were ultimately diagnosed with the condition compared with standard care alone.

These findings suggest that wearable devices may play a significant role in uncovering otherwise silent arrhythmias in older adults at increased stroke risk.

Clinical and economic implications

Atrial fibrillation is a well-established risk factor for stroke. Early detection allows for timely intervention, including anticoagulation therapy where appropriate, which can substantially reduce the likelihood of stroke.

Cardiologist Michael Winter of Amsterdam University Medical Center advocated for the integration of smartwatch technology into clinical pathways. He argued that the financial benefits could outweigh the initial expense of the devices.

He stated that the savings in medical services would “offset the initial cost of the device”.

Winter further explained:

“Using smartwatches with PPG and ECG functions aids doctors in diagnosing individuals unaware of their arrhythmia, thereby expediting the diagnostic process,” said Winter. “Our findings suggest a potential reduction in the risk of stroke, benefiting both patients and the healthcare system by reducing costs.”

By accelerating diagnosis in people who might otherwise remain undiagnosed until a serious event occurs, smartwatch-assisted monitoring may reduce both clinical burden and long-term healthcare expenditure.

Wearables as an adjunct – not a replacement

Despite these promising findings, the Apple Watch remains a supplementary tool rather than a replacement for clinical evaluation. Alerts generated by the device require follow-up assessment by healthcare professionals to confirm diagnosis and determine appropriate management.

However, for older adults at elevated risk of stroke, especially those who may not experience noticeable symptoms, wearable ECG and photoplethysmography technology may provide an additional layer of protection.

As consumer health technology continues to evolve, research such as this indicates that smartwatches could increasingly bridge the gap between everyday life and preventive cardiovascular care, supporting earlier detection of conditions that might otherwise go unnoticed.

Read More

Camera Glasses and Biomarkers Aim to Transform How Diets Are Measured in Real Life

Key Takeaways:

• A UK-wide study is testing camera glasses alongside blood and urine biomarkers to produce more accurate dietary data than self-reported food diaries.
  
• Researchers say existing methods often fail to capture snacking, portion sizes, and mindless eating, limiting confidence in nutrition research.
  
• While the technology could improve objectivity, experts caution it may not be suitable for everyone, particularly people vulnerable to food anxiety or disordered eating.
  
  
  

A new approach to measuring what people really eat

A new UK trial is exploring whether wearable camera glasses, combined with biological markers, can provide a more reliable picture of what people eat and drink in their everyday lives. The study is led by the University of Reading and aims to address long-standing challenges in nutrition research, particularly the limitations of self-reported dietary data.

Registered nutritionist Christine Bailey explained that camera glasses build on existing clinical tools, such as food photographs used to support dietary assessment, by capturing intake automatically and in real time. According to the research team, this approach could significantly reduce reliance on memory and self-reporting, which are known to be imperfect.

Professor Julie Lovegrove, who is leading the trial, said: “Humans are not very reliable, especially when asked to remember snacking or portion sizes.”

The study is taking place against a backdrop of rising rates of excess weight in the UK. Analysis from the Health Foundation in 2025 indicated that more than 60 percent of UK adults are now classified as living with overweight or obesity, including around 28 percent living with obesity.

How the SODIAT-2 study works

The study, known as SODIAT-2, will recruit 133 adults from across the UK to take part in a five-week programme conducted entirely from their own homes.

For up to 12 days, participants will wear camera glasses that automatically take photographs of everything they eat and drink. During this period, they will also collect small blood and urine samples using easy-to-use kits that are returned by post for laboratory analysis. In addition, participants will complete short online questionnaires to report what they have eaten over recent days.

All participants will then follow a standardised test diet, consuming identical foods and drinks for three days. By combining wearable imagery, biological data, and self-reported information, the research team aims to identify the most accurate and practical way to study diets in real-world settings.

Why current dietary assessment methods fall short

One of the central problems in nutrition research is obtaining an accurate account of people’s habitual eating patterns. Dr Manfred Beckmann, lead principal investigator from the Department of Life Sciences at the Aberystwyth University, said: “One of the problems facing nutrition researchers is getting a true picture of people’s eating habits.”

Professor Lovegrove noted that current approaches typically rely on food diaries, questionnaires, and 24-hour dietary recalls. She described these tools as “not very reliable or accurate”, largely because they depend on memory and honest reporting.

Christine Bailey added: “Research consistently shows that self-reported food diaries are prone to recall bias, with people often misremembering what they ate, when they ate it, and portion sizes.”

Dr Michelle Weech, research fellow at the University of Reading and trial manager, said: “By automatically photographing everything they eat and drink and measuring substances the body makes from food in their blood and urine – we will have dietary data we can really rely on.”

Potential benefits for nutrition and public health research

Researchers believe that combining camera glasses with biomarkers could mark a step change in how diets are measured. More accurate dietary data would allow scientists to explore links between diet, health, and disease with greater confidence, including conditions such as type 2 diabetes, cardiovascular disease, and some cancers.

Bailey said wearable camera technology may “improve objectivity and offer valuable insight into eating behaviours and patterns” that are not always captured through written food records alone.

Registered nutritional therapist Gemma Westfold, based in Windsor, highlighted how the technology could also shed light on behavioural aspects of eating. She said: “Humans can eat mindlessly on occasion, whilst scrolling on social media or while watching TV. When we are absorbed in other activities whilst eating, we can risk overeating, but more importantly, we can also switch off our ability to adequately digest and absorb the nutrients.”

Westfold added that using camera glasses for a short period could help identify behavioural patterns and support mindful eating, which she described as “key for all health conditions”.

Ethical considerations and concerns about over-monitoring

Despite the potential benefits, experts emphasise that such tools will not be appropriate for everyone. Bailey cautioned that increased monitoring could be counterproductive for some people: “For a small proportion of individuals, particularly those vulnerable to food anxiety or disordered eating, increased focus on food monitoring or quantity can become counterproductive and heighten preoccupation around eating.”

Westfold also raised concerns about how constant monitoring might affect therapeutic relationships. She said: “My line of work is based on relationships and making someone feel comfortable. Camera glasses could imply that as a nutritionist I do not trust my client, or believe what they are telling me.”

“It could make us feel like a food nanny, policing our clients and damage that relationship because they are under constant surveillance,” she added.

A collaborative UK research effort

The SODIAT-2 project brings together expertise from several UK institutions. Alongside the University of Reading and Aberystwyth University, partners include the University of Cambridge, which is leading blood sample analysis, and Imperial College London, which developed the camera glasses and is using artificial intelligence to analyse the images captured by the wearable devices.

Funded by the UK Medical Research Council and the Biotechnology and Biological Sciences Research Council, the project aims to improve how dietary intake is measured, providing stronger evidence to inform future nutrition guidance and public health policy.

Read More

NHS Endorses AI Notetaking to Expand Face-to-Face Patient Care

Key Takeaways: 

• NHS-backed AI notetaking tools could enable clinicians to spend up to a quarter more time with patients by reducing administrative burden.
  
• A new national registry of approved suppliers sets standards for clinical safety, technology assurance and data protection.
  
• Evidence from more than 17,000 patient encounters shows increased direct patient interaction and shorter appointment times when AI-scribing is used.
  
  
  

NHS support for ambient voice technologies

New artificial intelligence notetaking tools supported by the NHS could allow doctors to spend up to a quarter more time with people receiving care. NHS organisations across England are being encouraged to make use of a newly published national registry of approved suppliers offering this technology.

Often referred to as ambient voice technologies, these tools capture clinician–patient conversations and use AI to generate real-time transcriptions and clinical summaries. The aim is to reduce the time clinicians spend typing notes or navigating screens during consultations, while maintaining high standards of accuracy, privacy and data protection.

By adopting these systems, clinicians could save approximately two to three minutes per patient consultation. At scale, this time saving could be redirected towards additional appointments or more in-depth conversations with people seeking care.

New national registry sets standards for safety and data protection

NHS England has published a new self-certified registry for AI notetaking technologies, listing 19 suppliers that meet national requirements. The registry requires participating suppliers to comply with established standards covering clinical safety, technological performance and data protection.

The launch follows NHS guidance issued last year, which advised NHS organisations to adopt AI notetaking tools only where they are safe, evidence-based and demonstrably beneficial for patients. The registry is intended to give local NHS teams confidence when selecting and implementing these tools.

Clinical leadership highlights potential benefits

Dr Alec Price-Forbes, NHS England National Chief Clinical Information Officer, said:
“The AI revolution is here and we want to arm our NHS staff with the latest technology, which has the potential to transform the quality, safety and experience of care patients receive, as well as improving efficiency.

“AI notetaking tools will help free up more time for clinicians to focus on their patients, rather than typing up notes or looking at a screen – enhancing the quality of consultations and improving overall patient satisfaction.

“We are working with NHS organisations to help them implement the technology safely and effectively – helping to make the NHS the most AI-enabled healthcare system in the world, as we shift from analogue to digital.”

Minister for Digital Government Ian Murray also emphasised the wider public sector impact, stating:
“AI has enormous potential to transform public services, and this is a prime example of how we can use it to make a real difference. By cutting down on admin and paperwork, we’re giving clinicians back valuable time to do what they do best – caring for patients.

“We’re committed to making the UK an exemplar for how technology can be used to improve public services. Supporting the NHS to adopt tools like these safely and effectively is a key part of that mission.”

Evidence from NHS pilots and large-scale evaluation

AI notetaking technology has already been tested across nine NHS sites, where it was shown to free up clinicians to spend nearly a quarter more time with patients. A major NHS England-sponsored study published last year found that AI-scribing technology can significantly reduce clinician workload while supporting improvements in patient care. The findings suggest that national adoption could unlock millions of pounds worth of additional clinical activity.

The study was led by Great Ormond Street Hospital for Children NHS Foundation Trust Innovation Unit, known as GOSH DRIVE. It assessed the impact of an AI-scribing tool that automatically transcribes consultations and drafts summarised clinical notes for clinicians to review and approve.

Measurable improvements across care settings

More than 17,000 patient encounters were evaluated across a wide range of NHS settings, including hospitals, GP practices, mental health services and ambulance teams. The results demonstrated a 23.5 per cent increase in direct patient interaction time during appointments when AI-scribes were used. In addition, overall appointment length fell by 8.2 per cent.

Emergency departments saw particularly notable benefits, with a 13.4 per cent increase in the number of patients seen per shift. Together, these findings indicate that AI notetaking tools have the potential to improve both the experience of people receiving care and the efficiency of clinical services when implemented safely and appropriately.

Read More

AI-Enabled Digital Stethoscope Doubles Detection of Serious Valve Disease in Primary Care Study

Key Takeaways:

• An AI-enabled digital stethoscope more than doubled the sensitivity of detecting audible valvular heart disease compared with standard auscultation in primary care.
  
• The technology identified twice as many previously undiagnosed cases of moderate-to-severe disease, supporting its potential role as a screening adjunct.
  
• Higher sensitivity came with lower specificity, raising important considerations around false positives, referral rates, and cost-effectiveness.
  
  
  

Overview of the study

In a recent prospective study published in the European Heart Journal Digital Health, researchers compared the diagnostic accuracy of primary care providers using conventional stethoscopes with that of a relatively novel artificial intelligence-enabled digital stethoscope. The aim was to determine whether AI-supported auscultation could improve current approaches to identifying valvular heart disease in primary care settings.

The findings showed a marked improvement in sensitivity when AI support was used. The AI system demonstrated a sensitivity of 92.3 percent for detecting audible valvular heart disease, compared with 46.2 percent for standard care (P = 0.01). Although the AI tool showed slightly lower specificity, it identified twice as many cases of previously undiagnosed moderate-to-severe disease. This pattern suggests a potential role for AI-enabled auscultation as a screening adjunct rather than a replacement for clinical judgement and assessment.

Background

Valvular heart disease is a serious cardiac condition in which one or more of the heart valves, including the aortic, mitral, tricuspid, or pulmonary valves, fail to open or close properly, disrupting normal blood flow through the heart.

People living with valvular heart disease may experience symptoms such as shortness of breath, fatigue, chest pain, and palpitations. Prevalence increases with age and is estimated to affect more than half of adults aged over 65 to some degree, although moderate-to-severe disease is considerably less common.

Diagnosis remains challenging, in part because more than half of people with clinically significant disease are asymptomatic. Traditionally, detection relies on clinician-performed cardiac auscultation. However, previous research indicates that even experienced general practitioners may have limited sensitivity when screening asymptomatic individuals, contributing to delayed diagnosis and disease progression.

Study design and methods

The study investigated whether deep learning algorithms, combined with digital acoustic recordings, could improve the detection of cardiac abnormalities that may be missed during routine examinations.

This was a prospective, single-arm diagnostic accuracy study conducted across three primary care clinics between June 2021 and May 2023. The study included 357 participants aged 50 years and older who were considered at elevated cardiovascular risk but had no prior diagnosis of valvular heart disease or a known cardiac murmur.

Risk factors included hypertension, a body mass index of 30 or higher, diabetes, hyperlipidaemia, atrial fibrillation, previous myocardial infarction, stroke or transient ischaemic attack, coronary revascularisation, or other established cardiovascular disease.

Each participant underwent two independent screening protocols:

• Standard-of-care screening: Primary care providers performed four-point cardiac auscultation using conventional stethoscopes.
  
  
• AI-augmented screening: Study coordinators recorded phonocardiogram data using a digital stethoscope. These recordings were analysed by an AI algorithm that has received clearance from the US Food and Drug Administration to detect heart murmurs.
  
  

All participants subsequently underwent echocardiography to confirm the presence or absence of structural heart disease. An independent expert panel reviewed the digital audio recordings to verify whether an audible murmur was present. This panel was blinded to the AI results.

For the purposes of the study, audible valvular heart disease was defined as moderate-to-severe disease confirmed on echocardiography together with an expert-confirmed audible murmur. This definition acknowledged that some people with structurally significant disease may not produce a clearly audible murmur.

Study findings

The AI-augmented system substantially outperformed standard auscultation in detecting audible valvular heart disease. Sensitivity was 92.3 percent with AI support compared with 46.2 percent using standard-of-care screening (P = 0.01).

Among people with confirmed disease, standard examination missed seven of thirteen cases, whereas the AI system missed only one. In terms of previously undiagnosed moderate-to-severe valvular heart disease, the AI tool identified 12 cases, compared with 6 detected by primary care providers.

This improvement in sensitivity was accompanied by reduced specificity. The AI system demonstrated a specificity of 86.9 percent, compared with 95.6 percent for clinicians using conventional auscultation (P < 0.001), resulting in a higher number of false-positive findings.

When echocardiography alone was used as the reference standard for moderate-to-severe disease, regardless of whether a murmur was audible, the AI system continued to outperform standard care. Sensitivity in this analysis was 39.7 percent for the AI system versus 13.8 percent for clinicians (P = 0.01).

Interpretation and conclusions

The findings suggest that integrating AI-enabled digital stethoscopes into primary care could substantially improve the detection of valvular heart disease compared with traditional auscultation alone. Rather than replacing clinical assessment, these tools may provide an additional layer of screening support, helping clinicians identify people who may benefit from earlier referral and further investigation.

However, improved detection does not automatically translate into better clinical outcomes. The study assessed diagnostic accuracy but did not evaluate downstream management, patient experience, or long-term prognosis.

Several authors reported affiliations with the device manufacturer, a factor that should be considered when interpreting the results, despite transparent disclosure of conflicts of interest.

The lower specificity observed with AI-augmented screening may lead to increased referrals for echocardiography and higher healthcare utilisation. This highlights the importance of future research examining cost-effectiveness, workflow impact, and optimal integration into primary care pathways.

Study limitations included a modest sample size, a limited geographic scope, incomplete demographic detail, and the absence of systematic symptom assessment. Despite these constraints, the results indicate that AI-supported auscultation may represent a meaningful advance in point-of-care cardiac screening for people at increased cardiovascular risk.

Read More

AI-Enabled Social Robots Show Early Promise for Patient and Clinician Acceptance

Key Takeaways:

• A pilot study suggests that a GPT-controlled social robot is acceptable to both patients and healthcare professionals in a hospital setting.
  
• The research focused on technical, organisational and ethical feasibility, rather than on demonstrating improvements in care quality.
  
• Careful system design, including restricting information sources to clinician-validated content, was central to building trust and reducing risk.
  
  
  

Early insights into acceptance and feasibility

Researchers from University of Twente, Medisch Spectrum Twente and Politecnico di Milano have conducted a pilot study examining whether a GPT-controlled social robot could support people receiving care with medical information in a hospital environment. The initial findings suggest cautious optimism. Both patients and caregivers found the technology acceptable in practice.

The study examined not whether such a system improves clinical outcomes, but whether it can function safely and appropriately within real healthcare settings. Technical robustness, organisational fit and ethical considerations were all central to the research design.

Healthcare systems are facing sustained pressure from workforce shortages and rising demand. At the same time, clear, accessible communication remains essential, particularly for people living with chronic conditions. Digital tools may help address these challenges, but they also raise important questions around reliability, trust and governance.

The findings have been published in the journal Frontiers in Digital Health.

Exploring artificial intelligence with a physical presence

Within this context, the research team investigated whether a social robot, powered by GPT technology, could provide people receiving care with information about their condition and treatment. The system combined a physical robot with a human-like face, facial expressions and speech capabilities, enabling natural spoken interaction.

According to the study, this physical presence was well received by both patients and healthcare professionals. People described the conversations as accessible and pleasant. However, the researchers were careful to frame these findings appropriately.

“This should not be interpreted as evidence that care quality improves,” emphasised lead researcher Jan-Willem van ‘t Klooster. “We investigated whether such a system can function in practice, not whether it already improves care.”

Tested in real clinical settings

The research began with a controlled laboratory study before moving into everyday clinical practice. In total, 21 people with osteoarthritis and seven healthcare professionals interacted with the robot in the hospital setting. Both groups rated the system positively in terms of usability and overall acceptance.

Van ’t Klooster highlighted the importance of this early step. “Acceptance is a first step. Then you can investigate whether such a technology really contributes to better information provision, therapy adherence or time savings for health care providers.”

Managing risk through controlled use of AI

A key aspect of the project was how artificial intelligence was implemented. The GPT system did not have unrestricted access to the internet. Instead, it was limited to information drawn from pre-approved, clinician-validated medical websites. This approach was designed to reduce the risk of incorrect or fabricated responses, often referred to as hallucinations.

“The debate is often about whether you should use AI in health care,” said Van ’t Klooster. “We show that it is mainly about how you set it up. By setting clear boundaries, control remains in the hands of health care professionals.”

Collaboration across disciplines

The project brought together expertise from behavioural science, clinical practice, design and technology. Alongside researchers from the University of Twente, healthcare professionals, designers and international partners contributed to the study.

“It is precisely this collaboration that makes this kind of research possible,” Van ’t Klooster noted.

The authors stress that further work is needed before such systems could be considered for broader implementation. Planned follow-up research includes examining long-term use, knowledge transfer and the appropriate language level for patient communication, ensuring that future applications remain accessible, safe and trustworthy for people receiving care.

Read More

AI Tools Show Promise for Improving Diagnostics and Outcome Prediction in Resource-Limited Health Care Settings

Key Takeaways: 

• Transfer learning enables AI models trained in data-rich settings to be safely adapted for use in health care systems with limited local data, improving predictive performance without rebuilding models from scratch.
• In a Vietnam case study, an adapted AI model substantially improved prediction of neurological recovery after cardiac arrest compared with the unadapted model.
• Wider adoption of AI in low- and middle-income countries will require targeted skills development, infrastructure support and robust global governance frameworks.
  
  
  

Reducing uncertainty after cardiac arrest in constrained settings

After a cardiac arrest, families and clinicians are often confronted with profound uncertainty about a person’s chances of neurological recovery. This uncertainty is particularly acute in hospitals with limited resources, where access to advanced diagnostics and large datasets is constrained.

Researchers from Duke-NUS Medical School and collaborators have demonstrated how artificial intelligence can help address this challenge. By adapting an advanced AI model, the team improved the accuracy of neurological outcome prediction following cardiac arrest in a resource-limited setting.

Published in npj Digital Medicine, the study focused on the use of transfer learning, an AI technique that adapts models pre-trained on large datasets to new environments with limited local data. This approach can enhance performance in new contexts without requiring extensive and costly data collection, making it particularly relevant for low- and middle-income countries.

Adapting a high-income model for local use

The researchers began with a brain-recovery prediction model developed in Japan using data from 46,918 people who experienced out-of-hospital cardiac arrest. They then adapted this model for use in Vietnam, where it was tested on a smaller cohort of 243 patients.

The adapted model performed markedly better than the original model when applied directly to the Vietnamese data. It correctly distinguished between people at higher and lower risk of poor neurological outcomes approximately 80 percent of the time, compared with around 46 percent accuracy when the original, unadapted model was used.

Senior author Associate Professor Liu Nan, from Duke-NUS’ Center for Biomedical Data Science and Director of the Duke-NUS AI + Medical Sciences Initiative, said,

“The study shows AI models do not need to be rebuilt from scratch for every new setting. By adapting existing tools safely and effectively, transfer learning can lower costs, reduce development time and help extend the benefits of AI to health care systems with fewer resources.”

Expanding AI’s role beyond outcome prediction

Beyond predicting outcomes after cardiac arrest, AI has potential applications across a broad range of health care needs in low- and middle-income countries. In a separate study published in Nature Health, Duke-NUS researchers and collaborators, including colleagues from University College London, explored how large language models, trained on extensive text data to understand and generate human language, could support global health.

In resource-constrained environments, such tools may improve access to care, diagnostics and clinical decision-making. Examples highlighted by the researchers included a chatbot providing pregnancy-related information to expectant mothers in South Africa and smartphone-based applications used by community health workers in Sierra Leone to detect malaria infections from blood smear samples. These approaches offer more cost-efficient alternatives to conventional microscope-based systems.

Despite these advances, the researchers noted that AI development and deployment remain concentrated in high-income and upper-middle-income settings. While 63 percent of surveyed researchers, clinicians and service providers reported actively using AI tools, many low- and middle-income countries continue to face significant barriers, including limited infrastructure, insufficient technical expertise and a lack of locally generated evidence on how best to address these gaps.

Co-author Siegfried Wagner, from UCL Institute of Ophthalmology and Moorfields Eye Hospital NHS Foundation Trust, said,

“LLMs have the greatest opportunity to transform health care in settings where specialist physicians are scarcest, but the global health community needs to work together with some urgency to ensure the implementation of LLMs is supported in regions where adoption is most challenging.”

Dr Ning Yilin, Senior Research Fellow at Duke-NUS’ Center for Biomedical Data Science and a co-first author of the study, emphasised the importance of prioritising people when integrating AI into health care:

“Strengthening digital literacy and building confidence in using these tools will ensure AI supports, rather than disrupts, the workforce. Tailored skills-development pathways can help under-resourced workers adapt and thrive, allowing AI to uplift and add value to clinical and administrative roles.”

Charting the path forward: Governance and guardrails

While AI tools have clear potential to improve health care delivery, the researchers stressed that appropriate governance frameworks are essential to ensure safe and ethical implementation. Existing regulations for medical technologies often do not adequately address AI-specific risks, such as data privacy concerns, model hallucinations or unclear accountability for deployment and oversight.

To help close these gaps, researchers led by Duke-NUS have proposed the creation of an international consortium known as the Partnership for Oversight, Leadership, and Accountability in Regulating Intelligent Systems-Generative Models in Medicine, or POLARIS-GM.

The consortium aims to develop actionable best-practice guidance for regulating emerging AI tools, monitoring their impact, establishing safety guardrails and adapting them for use in resource-limited settings. By bringing together health care leaders, regulators, ethicists and patient groups from around the world, POLARIS-GM plans to adopt a phased approach, beginning with a review of existing research before working towards global consensus on AI governance in health care.

Dr Jasmine Ong, from the Duke-NUS AI + Medical Sciences Initiative and a Principal Clinical Pharmacist at Singapore General Hospital, and first author of the correspondence published in Nature Medicine, said,

“With clear oversight and clearly defined guidelines, health care systems can confidently leverage AI’s many strengths to improve health outcomes while steering clear of potential pitfalls.

From policymakers to patient groups, all stakeholders have a crucial role to play in making this goal a reality.”

Read More