Reproduction of scientific knowledge and norms for explaining schizophrenia in Predictive Psychological AI

Abstract

AI-based technologies for prediction of psychological and psychiatric states through means of identification, classification, diagnosing, and treatment monitoring – what I broadly term ‘Predictive<br/>Psychological AI’ (PPAI) - are increasingly being developed and used within the mental health area. PPAI technologies have showed promising results in predicting, diagnosing, and monitoring data points associated with various psychiatric conditions. However, few authors have analysed the fundamental scientific knowledge and normative assumptions about mental disorder or deviance found in such technologies, or traced the medical-historical foundations of such knowledge and norms and linked these to the technologies. Using the case of schizophrenia as studied in a number of PPAI studies and tracing its various conceptualisations throughout medical history, the purpose of this paper is to demonstrate how PPAI studies reproduce existing knowledge and norms for investigating schizophrenia. Specifically, I argue that PPAI studies reproduce naturalistic, genetic-neurobiological etiologies for the condition even though the empirical foundation of such etiologies may be subject to doubt. Such naturalistic ambitions to explain psychiatric conditions like schizophrenia can be traced back to ancient Greece, through the establishment of the first psychiatric hospitals, and into the early symptom-based formulation of schizophrenia and related diagnoses. Today, this medical-historical naturalistic tradition informs the data points that PPAI technologies seek to capture in order to predict schizophrenia as well as symptom severity and relapse risk among patients with the condition. Alongside these naturalistic aspirations within medical history, psychologically and sociologically grounded perspectives and research on schizophrenia and other mental health conditions have also persisted. However, I contend that within PPAI technologies these perspectives are either interpreted in ways that reinforce genetic-biological explanations of schizophrenia or are excluded altogether as data parameters I conclude the paper by discussing whether PPAI technologies can yield novel, potentially transformative insights into psychiatric diagnoses such as schizophrenia. I contend that for such technologies to achieve this, they must succeed in capturing the contextual and intersubjective variables that also correlate with the development of schizophrenia, rather than limiting their scope to variables that predominantly reflect existing genetic-neurobiological and symptomatic dimensions. If PPAI technologies are to be truly transformative, I<br/>suggest, it may further be necessary for them to break entirely with existing diagnostic categories. This would perhaps allow the knowledge they generate to inform the development of wholly new conceptual frameworks for identifying mental distress or deviation.

Ähnliche Arbeiten