Simon Hartmann

Predictive modelling and machine learning in mental illness

Only 30% of patients identified as at high risk of a psychotic episode transition to first episode psychosis. Improved accuracy of prediction is required to efficiently and safely intervene to prevent or minimise the impact of psychosis. As part of PRE-EMPT, the CRE for PREdiction of Early Mental Disorder and Preventive Treatment, we have access to detailed national and international data sets to develop novel Bayesian and machine learning techniques to combine clinical and biological variables to improve prediction accuracy.

Potential projects available for: Honours / PhD / Masters / Mphil

Speech and video biomarker extraction for mental health monitoring

The assessment of a patient’s mental health poses a complex challenge to clinicians. Structured interviews or questionnaires capturing a patient’s state are infrequently used in clinical practice resulting in a lack of standardised or systematically recorded data in mental health care. Hence, there is a need for objective measures that can be useful to identify mental illnesses. Due to the shift to online mental health assessment during Covid-19, there is increasing potential to facilitate care via extraction of video and speech features. This project aims to implement automated video and speech features extraction and processing to provide cross-sectional diagnostic and prognostic information.

Potential projects available for: Honours / PhD / Masters / Mphil

Big data analysis of cyclic alternating pattern (CAP) using machine learning

With the surge in wearable devices in recent years, the topic of what is high-quality sleep, how can it be determined and how can it be achieved attracted increasing interest. In the last two decades, cyclic alternating pattern (CAP) was introduced as a scoring alternative to traditional sleep staging. CAP is known as a synonym for sleep microstructure and describes oscillating brain waves defined as short EEG amplitude increases (<60 s) during NREM stages that are in tune with the rest of the body. In collaboration with leading research laboratories all over the world, we work on developing an automated CAP scoring algorithm which can be applied on large population based studies to investigate the role of CAP.

Potential projects available for: Honours