Toward the Use of Research and Diagnostic Algorithmic Rules to Assess the Recurrence of Illness and Major Dysmood Disorder Features: The Diagnosis “Bipolar Disorder” Is Useless

The DSM/ICD mood disorder classifications are incorrect, and their dogmatic nature prevents both inductive and deductive remodeling of diagnostic criteria. Using machine learning, we (a) proposed new Research and Diagnostic Algorithmic Rules (RADAR) to score lifetime (LT), current suicidal ideation and attempts, recurrence of illness (ROI), and the phenome of mood disorders; (b) developed precision models of mood disorders based on ROI and the phenome, as well as immune, nitro-oxidative, and gut–autoimmune pathways; and (c) built a new endophenotype "major dysmood disorder (MDMD)," which is characterized by increased ROI, a severe phenome, and the above pathways. The current narrative review demonstrates that the diagnoses of major (MDD) and bipolar (BD) disorder and bipolar 1 (BP1) and BP2 are ineffective because (a) a common core underpins recurrence of depressive and manic episodes, indicating that they are manifestations of ROI; (b) patients with MDD, BP1, and BP2 can be classified as MDMD or non-MDMD, indicating that MDD/BD/BP1/BP2 do not reflect overall severity; (c) because ROI is a more important indicator in our precision models, the binary diagnoses of MDD/BD/BP1/BP2 are meaningless; and (d) there are no significant differences in the model parameters concerning ROI and the phenome between unipolar and BD as found using measurement invariance assessment of composite models (MICOM) and multi-group analysis (MGA). In conclusion, diagnoses such as MDD/BD/BP1/BP2 should be abandoned in favor of our RADAR scores and the MDMD case definitions. Future studies should investigate which pathways underlie MDD + BD combined, ROI, and the first (hypo)manic episode.