The treatment of psychotic disorders uses trial and error, which means it often takes years to find the right treatment for patients, exposes them to side-effects and impairs recovery. Thus, there is a major clinical need for personalised treatment selection. Prior work by us and others showed that particular brain systems underlie symptoms in patients who do well to specific classes of treatment. We have developed imaging-based and other approaches to measure these and identified dopaminergic and glutamatergic circuits underlying treatment response. However, it remains unclear how best to combine these to inform treatment choices.
We have previously developed machine learning (ML) strategies that optimally combine neuroimaging and non-neuroimaging-based data modalities to produce superior prediction performance across different endpoints such as disease transition, functional impairment, or non-remitting formal thought disorder.
This PhD project will leverage these ML approaches to develop translational tools that can ultimately improve real-world patient care. Aims 1: determine the optimum combination of measures to predict treatment response in existing data; Aim 2: test models in independent data; Aim 3: Identify biological pathways underlying treatment non-response to help develop novel treatments by applying genetic and receptor pathway analysis to neuroimaging data. The PhD will give advanced training in ML, the use of advanced neuroimaging (eg fMRI, PET) and other biological techniques to measure brain systems, and training in clinical measures.
Objectives:
Yr 1: 1. Harmonise and integrate existing datasets; 2. apply machine learning techniques to identify the optimum approach to predict treatment response; 3. Use transparent ML methods to visualize response patterns at the individual-patient level,
Yr 2: 1. Acquire new data applying the best measures; 2. test the machine learning approaches in an independent dataset,
Yr 3: 1. Optimise the treatment prediction approach for clinical use; 2. Identify biological pathway(s) underlying non-response for novel treatments
Getting the right treatment for the right patient and novel treatment targets based on biology
Representative Publications
Veronese M, Santangelo B, Jauhar S, D’Ambrosio E, Demjaha A, Salimbeni H, Huajie J, McCrone P, Turkheimer F, Howes O. Neuropsychopharmacology. 2021 May;46(6):1122-1132.
Predicting treatment resistance from first-episode psychosis using routinely collected clinical information. Osimo EF, Perry BI, Mallikarjun P, Pritchard M, Lewis J, Katunda A, Murray GK, Perez J, Jones PB, Cardinal RN, Howes OD, Upthegrove R, Khandaker GM. NATURE Ment Health. 2023 Jan 19;1(1):25-35.
A neuroimaging biomarker for striatal dysfunction in schizophrenia. Li A, Zalesky A, Yue W, Howes O, Yan H, Liu Y, Fan L, Whitaker KJ, Xu K, Rao G, Li J, Liu S, Wang M, Sun Y, Song M, Li P, Chen J, Chen Y, Wang H, Liu W, Li Z, Yang Y, Guo H, Wan P, Lv L, Lu L, Yan J, Song Y, Wang H, Zhang H, Wu H, Ning Y, Du Y, Cheng Y, Xu J, Xu X, Zhang D, Wang X, Jiang T, Liu B. NATURE Med. 2020 Apr;26(4):558-565
Koutsouleris N, Dwyer D, Degenhardt F, …, Meisenzahl E, and the PRONIA Consortium. Multi-modal Workflows for Psychosis Prediction in Clinical High-Risk Syndromes and Recent-Onset Depression: A Multi-Site Machine Learning Analysis. JAMA Psychiatry. 2021; 78(2):195-209. doi: 10.1001/jamapsychiatry.2020.3604
Koutsouleris N, Kambeitz-Ilankovic L, Ruhrmann S, …, Borgwardt S, and the PRONIA Consortium. Individualized Prediction of Functional Outcomes in the Clinical High-Risk State for Psychosis and in Recent-Onset Depression: A Multi-modal, Multi-Site Machine Learning Analysis. JAMA Psychiatry. 2018; 75(11):1156-1172. doi: 10.1001/jamapsychiatry.2018.2165.
Koutsouleris N, Pantelis C, Velakoulis D, …, Schroeter M. Exploring links between Psychosis and Frontotemporal Dementia using Multi-Modal Machine Learning: ‘Dementia praecox‘ revisited. JAMA Psychiatry. 2022; 79(9):907-919. doi:10.1001/jamapsychiatry.2022.2075