Learning Hub
Sign up to our newsletter
20 July 2026
Funding from the King’s Health Partners Centre for Translational Medicine (CTM) has allowed Ninoslav Majkic to further his research on using AI to automate medication reviews for people with schizophrenia.
Please describe your day-to-day work and research interests
I am a senior pharmacist based at South London and Maudsley NHS Foundation Trust and a CTM Predoctoral Clinical Research Fellow in the Psychosis Studies Department at the Institute of Psychiatry, Psychology and Neuroscience (IoPPN).
My clinical work is based at the Trust’s Medicines Advice service, which provides expert advice on psychotropic medication use to several primary and secondary care services across the UK. My day-to-day work involves advising clinicians, service users, and carers on psychotropic medication use across a wide range of mental health conditions and comorbidities, including treatment-resistant disorders and rare diseases.
My research interests are in medicines optimisation in psychosis. Currently, my work is centred on exploring how electronic health records and digital health technology can be utilised to deliver personalised treatment for people with psychosis.
In particular, I am interested in how large language models (a recent development in artificial intelligence) can be used to automate antipsychotic medication reviews for people with schizophrenia. This work is supervised by Prof James MacCabe, Prof Richard Dobson, and Prof David Taylor.
What is the potential of your research to improve patient care?
Antipsychotics are the main treatment for schizophrenia, a debilitating disorder that carries a significant reduction in quality of life and life expectancy. Antipsychotic medication reviews help to optimise treatment to ensure the patient’s symptoms are adequately treated.
Since much information on past treatment with antipsychotics is recorded in unstructured, free-text entries in health records, medication reviews are time-consuming and as a result many patients will not have a review completed regularly.
My research hopes to address this time burden by using large language models to extract information relevant to antipsychotic medication reviews, to support clinicians to make an informed choice when prescribing treatment.
This research could ensure more patients have their antipsychotic medication optimised, minimise side-effect burdens from treatment, and improve earlier detection of treatment-resistant schizophrenia. This could hopefully reduce physical comorbidity, prevent relapse and admission to psychiatric units, and free up clinicians to spend more time with patients.
You received funding from the CTM, how has this helped your research?
The CTM predoctoral fellowship funding has been crucial for my research, providing me with protected time to carry out my research work. This has meant that I have been able to engage a patient and public involvement and engagement (PPIE) group and clinicians to support with the design of our project and begin developing our tool. It has also allowed me to present our design work at an international conference.
Crucially, I have been able to connect with other researchers with similar interests, and improve my understanding of the development and implementation of digital health technology, whilst providing me with space to develop my knowledge of research methodology.
How would you like to see your research progress?
We are currently developing the tool using our design work that has been completed so far. We will continue with analysis and evaluation of our tool, with the aim of implementing this into the clinical workflow.
By implementing this tool, we would hopefully pave the way for further development and implementation of applications that use artificial intelligence techniques into mental health care. Additionally, our tool has potential to be expanded to other medications and mental health conditions.
The next step will be to apply for larger funding opportunities to continue to explore how large language models can extract medication-related data in mental healthcare records.
This would help address an important gap in the clinical workflow, and could potentially aid mental health researchers to facilitate more effective identification or relevant data for research projects.
