Personalised use of medication in psychiatry with machine learning
How can we ensure better drug treatment in psychiatry from the first consultation? Tore Haslemo believes that machine learning is the answer.
In today's healthcare system, prescriptions for psychotropic drugs (psychiatric medicines) are often written in standard doses. It can take a lot of trial and error to find the right dose for the patient. Many factors, such as genetics, kidney function and other medications, play important roles in how much medicine is needed. In order to predict the dosage of psychotropic drugs prior to the start of treatment for each patient, Tore Haslemo and his team have utilised machine learning. A prototype has already been developed, and the goal is to create a decision support tool for doctors that will reduce medication errors and thus costs and side effects.
Presentation by Tore Haslemo, Head of the Centre for Psychopharmacology at Diakonhjemmet Hospital in Oslo.
The project is a collaboration between Diakonhjemmet Hospital and Deepinsight. Today's models can explain up to half of the variation in dose requirements between different patients, but the project wants to improve this figure even more. The machine learning model is under development. In addition, the project is looking at how the decision support tool can fit into the doctors' workflow.
Tore Haslemo is head of the Centre for Psychopharmacology at Diakonhjemmet Hospital in Oslo. He is also Associate Professor II at the pharmacy programme at OsloMet. His research interests include psychopharmaceuticals and genetics.
Recording
You can download the podcast to your mobile on Apple Podcasts, Spotify or Podbean. Search for ‘Norwegian Centre for E-health Research’.