Abstract
BACKGROUND: Fingerprint patterns are developed during pregnancy and share a common embryogenic origin with the central nervous system. Considering the observed relationship between prenatal abnormalities and higher risk for schizophrenia, we previously built fingerprint-based algorithms achieving validation accuracies up to 70% in a large sample of patients with schizophrenia. In this new study, we apply them to a sample of patients with bipolar disorder. METHODS: Besides validating the developed algorithms with an independent sample of fingerprints from N = 127 patients with schizophrenia and N = 116 healthy controls, here we applied them to a sample of N = 118 bipolar disorder patients. Scores from a premorbid IQ scale were also obtained from all participants, and the link between these scores and the algorithm outcomes was analyzed. RESULTS: The initial validation provided balanced accuracies similar to those of the original study (57%-68%). When applied to subjects with bipolar disorder (against healthy individuals), algorithms also showed significant predictive power (accuracies: 55%-68%). Consequently, the capacity to discriminate between schizophrenia and bipolar disorder was poor (accuracies: 47%-57%). Regression analyses between averaged probabilities and premorbid IQ scores were significant in schizophrenia (r = -0.184; p = 0.041) and in the whole sample (r = -0.159; p = 0.002). CONCLUSIONS: Algorithms were predictive of bipolarity, highlighting the existence of fingerprint abnormalities also in bipolar disorder. In addition, the observed associations with premorbid IQ underline the neurodevelopmental origin of fingerprint patterns and suggest the potential use of fingerprints for prediction in other neurodevelopmental disorders. The lack of specificity with the diagnosis of schizophrenia, though, points to the need for new algorithms for differential diagnosis in psychosis.