A TRUSTworthy speech-based AI monitorING system for the prediction of relapse in individuals with schizophrenia

UFUK 2020 Projesi, 2023 - 2029

Proje Türü: UFUK 2020 Projesi
Başlama Tarihi: Ekim 2023
Bitiş Tarihi: Mart 2029

Proje Özeti

Schizophrenia is a serious psychiatric disorder that often starts during adolescence and lasts a lifetime. It is characterized by psychosis, which typically has a waxing and waning course with remissions and relapses and is estimated to affect 21 million people worldwide. Schizophrenia contributes 13.4 (9.9–16.7) million years of life lived with disability (DALY) to the burden of disease - the highest DALY among mental and substance-use disorders and ranking 8th place among all diseases1,2 Relapse and associated hospital admissions can be deeply distressing and traumatic for those affected. A psychotic relapse is estimated to cause 9 months lived with disability, and the direct treatment costs are 3 times higher in case of relapse than for those with sustained remission3. Patients with a relapsing course have reduced chances of sustaining relationships, higher risks of unemployment and more severe functional and cognitive decline4. Maintenance treatment with antipsychotic medication is an effective method to prevent relapse5 but has serious side effects such as metabolic syndrome and parkinsonism, especially when used long term6,7. Over the last decade, society has witnessed a change in attitude and for many patients, maintenance treatment is no longer acceptable8. Many patients want to stop antipsychotic medication after remission, and 56% actually do so9,10. During and after medication discontinuation, relapse risk is very high and frequent monitoring becomes a necessity, which is not feasible with scarce clinical resources. Ground-breaking work from Dr. Spaniel (partner 7) on the temporal aspects of relapse, showed a gradual course of early signs over some 5 weeks5. He also showed that in these early stages, a relapse can be prevented effectively when an individualized relapse prevention plan is activated11. To accommodate this societal change in medication use, accurate and timely relapse prediction is crucial, as psychotic relapse can still be prevented using intermittent medication and/or psychosocial interventions, if such actions are taken in time12,13. This is a highly challenging task, given the heterogeneity in the clinical presentation of psychotic relapse and the inherent difficulty of monitoring citizens outside the clinic. Thus, there is an urgent need to develop valid and trustworthy predictors of relapse, but no such tools currently exist.