Application of network methods for understanding mental disorders: pitfalls and promise

Abstract

Galvanized with the availability of sophisticated statistical techniques and large datasets, network medicine has emerged as an active area of investigation. Following this trend, network methods have been utilized to understand the interplay between symptoms of mental disorders. This realistic approach that may provide an improved framework into understanding mental conditions and underlying mechanisms is certainly to be welcomed. However, we have noticed that symptom network studies tend to lose sight of the fundamentals, overlook major limitations embedded in study designs, and make inferences that are difficult to justify with current findings. There is concern that disregarding these flaws may halt the progress of the network approach in psychiatry. Therefore, in this paper, we first attempt to identify the pitfalls: (1) a reductionist understanding of medicine and psychiatry, thereby inadvertently reintroducing the dichotomy of medicine (lung cancer) and psychiatry (depression), (2) a shortsighted view of signs and symptoms, (3) overlooking the limitations of available datasets based on scales with embedded latent class structures, (4) overestimating the importance of the current findings beyond what is supported by the study design. By addressing current issues, the hope is to navigate this rapidly growing field to a more methodologically sound and reproducible path that will contribute to our understanding of mental disorders and its underlying mechanisms.