Researchers at the University of Melbourne have developed a groundbreaking AI tool, DT-GPT, which creates virtual patient representations and predicts individual health trajectories, particularly for those with Alzheimer’s or lung cancer. Trained on extensive electronic health records, the model forecasts health changes over time, leveraging detailed patient histories and lab results without prior knowledge of patient outcomes. Lead researcher Associate Professor Michael Menden highlighted its accuracy, outperforming 14 other models in predictive capabilities. DT-GPT also facilitates personalized medicine by enabling early intervention and customized treatment plans, enhancing patient care and health outcomes. Its conversational interface allows users to interact like a chatbot, while the model’s ability for ‘zero-shot predictions’ demonstrates its potential to anticipate untrained laboratory values accurately. Recently published in NPJ Digital Medicine, this technology has sparked the establishment of a new company focused on creating digital twins for endometriosis patients, underscoring its versatile applications in clinical trials and drug development.
Source link
Share
Read more