Skip to content

Research

Background

AI-based approaches such as machine learning using neural networks are increasingly being used in neuropsychiatric research and practice. They are expected to enable more precise prediction, early detection and diagnosis, and thus more targeted treatment of neuropsychiatric disorders. However, the basis of the results obtained with these computational psychiatry methods is often no longer readily comprehensible to professional users as well as to patients and their relatives. This opacity of AI-based approaches raises fundamental theoretical, ethical, and social questions for neuropsychiatric research and practice: How are AI-based expert systems changing our understanding of psychiatry and neuropsychiatric disorders? What effects do they have on the role of the physician, the identity of patients, and the relationship between the two?

The joint project TESIComP investigates these questions with the help of different methods of qualitative social research as well as ethical and scientific analysis and is divided into three scientific subprojects:

Subproject 1: Ethical and Philosophy of Science Analysis

Subproject 1 is led at Carl von Ossietzky University by Prof. Dr. Mark Schweda:

SP1 is dedicated to empirically informed ethical and scientific theoretical analysis of the significance of the development of computational psychiatry approaches. The focus is on the implications of AI-based approaches for psychiatric research and clinical practice. In doing so, we address three guiding questions: (1) How does the development of increasingly precise algorithm-based disease categories and treatment regimens that can no longer be explained or understood in the traditional sense change the traditional understanding of neuropsychiatric disorders, which is oriented toward the interpretation of clinical symptoms, and thus the self-understanding of psychiatry as a scientific discipline? (2) What are the implications of AI-based procedures for neuropsychiatric diagnosis and treatment decisions for the professional self-image and ethos of psychiatrists (medical authority, judgment, and responsibility), the identity and social role of psychiatric patients (self-interpretation, labeling, (de-)stigmatization), and the relationship, interaction, and communication between both sides (trust, informed consent, shared decision making)? (3) How should the future development of computational psychiatry be assessed scientifically, technically, and in terms of medical and health care ethics, and how should it be embedded in a socially and socio-politically meaningful way (identification of relevant topics and challenges for public and political debate, need for regulation, e.g. with regard to questions of liability and solidarity-based cost assumption)?

Methodology

Theory-building project phase:

Exploratory project phase:

Evaluative project phase:

Subproject 2: Empirical exploration in dementia practice.

Subproject 2 is led by Prof. Dr. Stefan Teipel at the DZNE in the Helmholtz Gesellschaft at the Rostock/Greifswald site:

The focus of TP2 is on the implications of computational psychiatry and the use of AI-based approaches in the context of neuropsychiatric diagnosis and treatment of dementia. The guiding question aims to address how the development of algorithm-based precision diagnostics is changing the understanding and management of dementia at the scientific and clinical-practical level. A concrete use case is the explaining prototype of an AI diagnostic system (https://explaination.net/demo).

The impact of AI-assisted approaches in the context of dementia will be explored in four respects: (1) regarding the self-concept of psychiatry and its concept of neurodegenerative brain diseases. The use of biomarkers has already initiated a biological turn here, which largely detaches from a symptom-based diagnosis and thus at the same time runs the risk of losing sight of the patient perspective. New AI-supported systems have the potential to accelerate and reinforce this development; (2) with regard to psychiatrists and other specialists involved, whose professional self-image is being changed by biological explanatory models. The question is to what extent treating physicians can combine an AI-based disease diagnosis with an understanding perspective of the patients regarding their disease. In this context, questions about the attribution of responsibility in the case of misdiagnosis will also be investigated, which AI systems pose anew and which place the explainability and comprehensibility of the systems by treating physicians in the foreground. (3) Concerning the self-understanding of patients, who can explain their complaints with brain-organic disease changes visualized by means of AI, but who also run the risk that their own experience of the disease takes a back seat to diagnostics and care. (4) Regarding the doctor-patient relationship that is altered by the use of AI systems in diagnostics. To investigate whether such a system broadens or narrows communication about the disease; that is, does the patient’s experience remain the focus of the doctor-patient conversation or is it replaced by a purely biological interpretation of the disease and its mediation.

Methodology:

Theory-building project phase:

Exploratory project phase:

Evaluative project phase:

Subproject 3: Empirical exploration in the practice of depression.

Subproject 3 is led at the university by Prof. Dr. Oliver Gruber:

TP3 is dedicated to the importance of computational psychiatry and in particular AI-based approaches in the context of depression. The aim is to investigate how the development of algorithm-based precision diagnostics and corresponding individualized therapies influences the understanding and management of depressive disorders in neuropsychiatric research and practice.

The empirical exploration of this guiding question will be conducted from three different perspectives: First, from a scientific theoretical perspective, it will be analyzed to what extent the use of AI changes the self-understanding and disease concepts of psychiatry with regard to depressive disorders. Second, from the perspective of the treating psychiatrists, it will be explored to what extent their professional self-image is affected by the use of AI-supported approaches. In this context, questions about the attribution of responsibility in the case of misdiagnosis, which can occur here in a new form, should also be investigated. Third, from the patients’ perspective, the impact of AI-assisted diagnosis on their clinical care and their personal self-image as depressed individuals will be analyzed. In addition, the role of both the practitioner and the treated will be explored in more detail in the course of considering the influence of AI-based approaches on the relationship, interaction, and communication between practitioner and patient.

Methodology:

Theory-building project phase:

Exploratory project phase:

Evaluative project phase: