The Characteristics of Treatment Resistant Schizophrenia From the Illness Onset

Previous long-term follow-up studies on patients with first-episode schizophrenia have shown that up to 30% of patients who have never received antipsychotic medication treatment do not experience symptom relief or have poor treatment response after standard antipsychotic medication treatment, becoming treatment-resistant schizophrenia (TRS). Moreover, in long-term follow-up, patients with treatment-resistant schizophrenia from the illness onset (TRO) account for 80% of all TRS patients. Preliminary studies abroad have found that TRO patients have characteristics such as early age of onset, male predominance, prominent negative symptoms, high proportion of positive family history, and long duration of untreated psychosis, but there is still no consistent conclusion on the pathological mechanisms. There is currently no research on this type of patient in China, and there are difficulties in early diagnosis of TRO patients in clinical practice. This study aims to establish a TRO prediction model by integrating data on demographics, disease characteristics, psychopathology, social function, and neurocognition from a cohort of patients with first-episode schizophrenia. Mathematical modeling methods such as K-Means/SVM and convolutional neural networks will be used. Therefore, in patients with untreated first-episode schizophrenia, early and accurate identification of TRO patients at the initial diagnosis stage and treatment with clozapine is particularly important for potentially shortening the treatment period and reducing the personal and societal burden of TRO patients. Based on the progress of existing research and the previous work of the research team, we speculate that TRO patients have unique clinical features. This project will establish a TRO prediction model based on multidimensional clinical data using mathematical modeling methods. From a clinical application perspective, the study selects TRO model prediction factors based on existing clinical assessment methods, making the model highly clinically applicable and generalizable. By establishing a TRO prediction model, not only can high-risk TRO patients be identified early in the initial diagnosis stage, enabling appropriate clinical treatment interventions, but it can also provide new insights into the future clinical treatment of TRO, promote the development of early and personalized precision identification and treatment of TRO, and improve long-term prognosis and reduce the burden of the disease for patients.