Scipher Lab
Schizophrenia Causal Inference in Physical Health Evaluation Research
We develop causal and actionable prediction models using counterfactual explanations to understand and predict physical health outcomes in patients with psychosis.
Schizophrenia Causal Inference in Physical Health Evaluation Research
We develop causal and actionable prediction models using counterfactual explanations to understand and predict physical health outcomes in patients with psychosis.
Scipher Lab is located in the Department of Psychiatry at the University of Cambridge. We are a multidisciplinary research group focused on understanding and predicting physical health outcomes in individuals with psychosis.
Making informed clinical decisions based on individualised outcome predictions is the cornerstone of precision psychiatry. Prediction models currently employed in psychiatry rely on algorithms that map a statistical relationship between clinical features (predictors/risk factors) and subsequent clinical outcomes. They rely on associations that overlook the underlying causal structures within the data. Such predictions from sparse associative models from routinely collected data are rarely actionable at an individual level. To be actionable, prediction models should address these shortcomings.
We believe that precision psychiatry requires not just accurate predictions, but explanations that clinicians can act upon. The fundamental goal of prediction in precision psychiatry is not to just generate risk probabilities; it is to improve individual patient outcomes through informed clinical decision-making. A prediction model that cannot answer the causal question -"What is the optimum intervention for this specific patient / Will intervening on this specific factor improve this specific patient's outcome?" - has fundamentally failed at the task precision medicine demands, regardless of how accurately it predicts. This is where our CAPE (Counterfactual Analysis for Prediction Explanation) framework comes in.
Causal and Actionable Prediction Models: Using counterfactual explanations to understand which clinical interventions could modify predicted outcomes in physical health across the psychosis spectrum.
Neuroimaging • Machine Learning • Causal Inference • Clinical Prediction • Remote Monitoring / Virtual Wards • Clozapine • Physical Health in Psychosis • Schizophrenia
We focus on multiple complementary research streams to develop comprehensive understanding of physical health in psychosis.
Developing causal structural equation models and directed acyclic graphs (DAGs) to understand the relationships between clinical variables and physical health outcomes.
Creating interpretable explanations for predictions by identifying what-if scenarios—which interventions or modifications to clinical variables would change predicted outcomes.
Applying supervised and unsupervised learning techniques to large datasets to identify patterns predictive of physical health outcomes and treatment response.
Advancing precision psychiatry by moving beyond simple predictions to actionable, clinically relevant recommendations tailored to individual patient profiles.
Leveraging digital technology for community-based care. Dr Krishnadas runs the innovative Peterborough clozapine clinic, featuring one of the UK's first virtual wards for community clozapine initiation. Our team won the RCPsych Digital Mental Health Team of the Year award in 2025.
Our diverse group of clinicians, researchers, and students work collaboratively on advancing precision psychiatry.
Scipher Lab is engaged in several funded research projects advancing precision psychiatry and causal inference in psychosis.
Our research has been published in leading peer-reviewed journals. View our full publication list on ORCID and Google Scholar.
A fallacy at the heart of precision medicine - are current clinical prediction models actionable?
Leighton, S., Deligianni, F., Tsaftaris, S., & Krishnadas, R. (2025)
DOI: 10.31234/osf.io/jw37c_v1Antipsychotic-induced weight gain in psychosis – a causal mediation analysis and feasibility study
Leighton, S., Leong, I. L., Machlanski, D., Perry, B., Tsaftaris, S., Deligianni, F., & Krishnadas, R. (2025)
DOI: 10.31234/osf.io/zfhx3_v3Ethnic Bias in Prediction and Decision Making Algorithms in Precision Psychiatry: Challenges in a Shrinking World
Krishnadas, R. Journal of Psychosocial Rehabilitation and Mental Health (2025)
DOI: 10.1007/s40737-025-00472-0Precision psychiatry: thinking beyond simple prediction models–enhancing causal predictions
Krishnadas, R., Leighton, S. P., Jones, P. B. The British Journal of Psychiatry (2025)
A key perspective article on moving beyond simple prediction models to actionable, causal insights in precision psychiatry.
Navigating the new frontier: psychiatrist's guide to using large language models in daily practice
Krishnadas, R. BJPsych Advances (2025)
DOI: 10.1192/bja.2025.10186Interested in joining the lab, collaborating, or learning more about our research? Get in touch.
Department of Psychiatry
18b Trumpington Road
Cambridge, CB2 8AH
United Kingdom
Principal Investigator:
Dr. Rajeev Krishnadas
rk758@cam.ac.uk