New AI Tool Predicts Medical Events to Support Clinical Decision-Making in Healthcare Settings

This innovative tool could be used to aid clinical decision-making, healthcare monitoring, and improving the efficiency of clinical trials.

The tool – called Foresight – was developed by a team of researchers that included investigators from King’s College London (London, UK) and trained on data from extensive NHS electronic health records (EHRs). Foresight uses a deep learning approach to recognize complex patterns within the vast data of EHRs, both structured and unstructured, for generating predictive insights. The team utilized data from more than 811,000 patients for training three distinct Foresight models and extracted and processed the unstructured (free-text) and structured data (age, ethnicity, and sex) within the EHRs.

The team validated its predictive accuracy by comparing how well its predictions matched the actual health outcomes noted in a smaller data subset. When forecasting the next 10 possible disorders that could appear next in a patient timeline, Foresight correctly identified the next disorder 68% and 76% of the time in two UK NHS Trusts and 88% of the time in the US MIMIC-III dataset. Similarly, when forecasting the next new biomedical concept which could be a disorder, symptom, relapse, or medication, Foresight achieved a precision of 80%, 81%, and 91%, respectively.

Clinicians also evaluated Foresight's accuracy by creating mock patient timelines with various medical scenarios. When a unanimous agreement on a predicted medical event was reached among the clinicians, Foresight's predictions were found to be 93% relevant from a clinical standpoint. This showcases Foresight's capability for practical applications in risk forecasting, clinical research emulation, disorder progression studies, intervention simulations, and educational purposes.

“Our study shows that Foresight can achieve high levels of precision in predicting health trajectories of patients, demonstrating it could be a valuable tool to aid decision making and inform clinical research,” said Zeljko Kraljevic, Research Fellow in Health Informatics at King's College London. “The proposed purpose of Foresight is not to enable patients to self-diagnose or predict their future, but it could potentially be used as an aid by clinicians to make sure a diagnosis is not missed or for continual patient monitoring for real-time risk prediction. One of the main advantages of Foresight that it can easily scale to more patients, hospital or disorders with minimal or no modifications, and the more data it receives the better it gets.”

Related Links:
King’s College London