Can I use machine learning to build a prediction model of individual patient outcome in first episode psychosis?

Can I use machine learning to build a prediction model of individual patient outcome in first episode psychosis?

Psychosis is an illness manifest by unusual or muddled thoughts and hearing voices. Psychosis is the fifth leading UK cause of disability among working age adults. Meaningful recovery is more than treating symptoms but includes positive quality of life, social and functional outcomes. At a group level, we know factors like duration of untreated psychosis are associated with worse outcomes. However, we struggle to predict who will do well at an individual level. An advanced statistical technique called machine learning has the potential to revolutionise medicine by the development of models which can predict outcome in individual patients. To date, few studies have looked at how machine learning models generalise to patients with psychosis across different clinical settings, geographical locations and time periods. Without this step, we cannot be sure that the models are accurate outside the original group of patients on which they were built.Can I use machine learning to predict individual outcome in psychosis? Do these models perform well in patients from different places and time periods? I hope my research will lead to a personalised approach to care, maximising available resources, with considerable benefit to patients.

[{ "PostingID": 4004, "Title": "LILLY-F1D-MC-HGDH", "Description": "The Acute and Long-Term Efficacy of Olanzapine in First-Episode Psychotic Disorders: A Randomized Double-Blind Comparison with Haloperidol" }]