AI could help predict medical conditions earlier thanks to unique health data sets
Nightingale Open Science, a project launched by Ziad Obermeyer, a physician and machine learning scientist at the University of California, Berkeley, offers a unique and free medical dataset aimed at solving medical mysteries through the use of artificial intelligence (AI).
The project, which was funded with $2 million from former Google CEO Eric Schmidt, consists of 40 terabytes of medical imagery such as X-rays, electrocardiogram waveforms, and pathology specimens from patients with various conditions, including Covid-19, sudden cardiac arrest, fractures, and high-risk breast cancer. Each image is labelled with the patient’s medical outcomes, such as the stage of breast cancer, whether it resulted in death, or whether a Covid-19 patient required a ventilator.
The datasets were curated with the help of hospitals in the US and Taiwan over a period of two years, and Obermeyer plans to expand this to include more medical diversity in Kenya and Lebanon in the coming months. The key differentiating factor between these datasets and those that are available online is that they are labelled with “ground truth,” meaning that they are based on what actually happened to the patient rather than a doctor’s opinion.
The AI community has recently shifted its focus from collecting “big data” to more meaningful data, which is relevant to a specific problem and can be used to address issues such as inherent biases in healthcare, image recognition, or natural language processing. In healthcare, algorithms have been shown to amplify existing health disparities. For example, an AI system used by hospitals to allocate additional medical support for patients with chronic illnesses was found to be prioritising healthier white patients over sicker black patients who needed help, due to the use of healthcare costs as a proxy for healthcare needs.
The Nightingale Open Science project is aimed at producing high-quality datasets that are more representative of the population and can be used to root out underlying biases that are discriminatory towards underserved and underrepresented groups in healthcare systems, such as women and minorities. These diverse datasets can also help to study the spread of diseases in different cultures and locations, where people may react differently to illnesses.
The availability of high-quality, diverse medical datasets can enable AI to predict medical conditions earlier, triage better, and ultimately save lives. The Nightingale Open Science project, which provides unique and curated datasets based on actual patient outcomes, has the potential to drive progress in healthcare and reduce the impact of biases on healthcare systems.