AI Predicts Future Health Outcomes of Patients: Study Finds

Recently, a remarkable study has found that artificial intelligence (AI) technology can accurately predict the types of health conditions that patients may face in the future, bringing innovative applications to the medical field.

According to reports, this AI tool called Foresight, belongs to the advanced AI model family along with ChatGPT, but what sets it apart is its use of rich information from NHS electronic records for training. This research was jointly developed by researchers from King's College London (KCL), University College London (UCL), and two well-known NHS trust organizations - King's College Hospital NHS Foundation Trust and Guy's and St Thomas' NHS Foundation Trust.

Researchers made full use of data from over 811,000 patients in the two NHS trust organizations in London, combined with publicly available datasets from the United States, to carefully train three different Foresight models. These models can predict 10 potential diseases that patients may face in the future based on their medical history.

Excitingly, Foresight demonstrated excellent predictive ability in the experiment. When using data from NHS hospitals, it was able to correctly identify conditions with an accuracy rate of 68% to 76%. When using US data, the accuracy rate reached an impressive 88%, fully demonstrating the tool's outstanding performance in predicting patients' health trajectories.

Zeljko Kraljevic, a researcher in Health Informatics, Biostatistics, and Health Informatics at KCL and the first author of the study, highly praised the predictive ability of Foresight. He stated that the research results fully demonstrate the high level of accuracy of the tool in predicting patients' health trajectories, making it a valuable tool for decision support and clinical research.

Richard Dobson, a senior author of the study and a professor of Medical Informatics at KCL and UCL, also expressed excitement about this achievement. He emphasized that this is an important breakthrough for artificial intelligence in healthcare and that appropriate data must be used for model training to ensure the effectiveness of the tool and work towards the common goal of supporting healthcare systems and benefiting patients.

Currently, the research team is actively seeking more participation from hospitals to further develop an upgraded version of Foresight 2. Professor Dobson revealed that this new version will be a more accurate language model, expected to bring more precise predictions and decision support to the medical field.