Artificial intelligence (AI) could predict the type of health conditions a patient is likely to develop in the future, a study has found.
Using the technology in this way could help doctors when it comes to monitoring patients or making decisions around diagnosis, researchers said.
The AI tool, known as Foresight, belongs to the same family of models as ChatGPT but has been trained using information from NHS electronic records.
It was developed by researchers from King’s College London (KCL), University College London (UCL), King’s College Hospital NHS Foundation Trust, and Guy’s and St Thomas’ NHS Foundation Trust in the U.K.
They trained three different models of Foresight using data from more than 811,000 patients from across two NHS trusts in London and a publicly available dataset in the U.S.
It was given 10 possible disorders a patient may experience next based on their records.
When using data from U.K. hospitals, Foresight was able to correctly identify the condition 68% and 76% of the time. It was accurate 88% of the time when using U.S. data.
Zeljko Kraljevic, a research fellow in health informatics, biostatistics and health informatics at KCL, developed Foresight and is the first author of the study.
He said the findings – published in The Lancet Digital Health – show the tool "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."
Kraljevic added: "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 is that it can easily scale to more patients, hospitals or disorders with minimal or no modifications, and the more data it receives the better it gets."
Foresight’s accuracy was also assessed by medics as part of the study.
Five clinicians developed 34 mock patient timelines with simulated scenarios. When they all agreed on a forecast medical event, Foresight’s predictions also made sense from a clinical perspective 93% of the time.