Evidence-based Shared-Decision-Making Assistant (SDM-assistant) for choosing antipsychotics: protocol of a cluster-randomized trial in hospitalized patients with schizophrenia.

BACKGROUND: Choosing an antipsychotic medication is an important medical decision in the treatment of schizophrenia. This decision requires risk-benefit assessments of antipsychotics, and thus, shared-decision making between physician and patients is strongly encouraged. Although the efficacy and side-effect profiles of antipsychotics are well-established, there is no clear framework for the communication of the evidence between physicians and patients. For this reason, we developed an evidence-based shared-decision making assistant (SDM-assistant) that presents high-quality evidence from network meta-analysis on the efficacy and side-effect profile of antipsychotics and can be used as a basis for shared-decision making between physicians and patients when selecting antipsychotic medications. METHODS: The planned matched-pair cluster-randomised trial will be conducted in acute psychiatric wards (n = 14 wards planned) and will include adult inpatients with schizophrenia or schizophrenia-like disorders (N = 252 participants planned). On the intervention wards, patients and their treating physicians will use the SDM-assistant, whenever a decision on choosing an antipsychotic is warranted. On the control wards, antipsychotics will be chosen according to treatment-as-usual. The primary outcome will be patients' perceived involvement in the decision-making during the inpatient stay as measured with the SDM-Q-9. We will also assess therapeutic alliance, symptom severity, side-effects, treatment satisfaction, adherence, quality of life, functioning and rehospitalizations as secondary outcomes. Outcomes could be analysed at discharge and at follow-up after three months from discharge. The analysis will be conducted per-protocol using mixed-effects linear regression models for continuous outcomes and logistic regression models using generalised estimating equations for dichotomous outcomes. Barriers and facilitators in the implementation of the intervention will also be examined using a qualitative content analysis. DISCUSSION: This is the first trial to examine a decision assistant specifically designed to facilitate shared-decision making for choosing antipsychotic medications, i.e., SDM-assistant, in acutely ill inpatients with schizophrenia. If the intervention can be successfully implemented, SDM-assistant could advance evidence-based medicine in schizophrenia by putting medical evidence on antipsychotics into the context of patient preferences and values. This could subsequently lead to a higher involvement of the patients in decision-making and better therapy decisions. TRIAL REGISTRATION: German Clinical Trials Register (ID: DRKS00027316 , registration date 26.01.2022).

Modeling aspects of the language of life through transfer-learning protein sequences.

BACKGROUND: Predicting protein function and structure from sequence is one important challenge for computational biology. For 26 years, most state-of-the-art approaches combined machine learning and evolutionary information. However, for some applications retrieving related proteins is becoming too time-consuming. Additionally, evolutionary information is less powerful for small families, e.g. for proteins from the Dark Proteome. Both these problems are addressed by the new methodology introduced here. RESULTS: We introduced a novel way to represent protein sequences as continuous vectors (embeddings) by using the language model ELMo taken from natural language processing. By modeling protein sequences, ELMo effectively captured the biophysical properties of the language of life from unlabeled big data (UniRef50). We refer to these new embeddings as SeqVec (Sequence-to-Vector) and demonstrate their effectiveness by training simple neural networks for two different tasks. At the per-residue level, secondary structure (Q3 = 79% ± 1, Q8 = 68% ± 1) and regions with intrinsic disorder (MCC = 0.59 ± 0.03) were predicted significantly better than through one-hot encoding or through Word2vec-like approaches. At the per-protein level, subcellular localization was predicted in ten classes (Q10 = 68% ± 1) and membrane-bound were distinguished from water-soluble proteins (Q2 = 87% ± 1). Although SeqVec embeddings generated the best predictions from single sequences, no solution improved over the best existing method using evolutionary information. Nevertheless, our approach improved over some popular methods using evolutionary information and for some proteins even did beat the best. Thus, they prove to condense the underlying principles of protein sequences. Overall, the important novelty is speed: where the lightning-fast HHblits needed on average about two minutes to generate the evolutionary information for a target protein, SeqVec created embeddings on average in 0.03 s. As this speed-up is independent of the size of growing sequence databases, SeqVec provides a highly scalable approach for the analysis of big data in proteomics, i.e. microbiome or metaproteome analysis. CONCLUSION: Transfer-learning succeeded to extract information from unlabeled sequence databases relevant for various protein prediction tasks. SeqVec modeled the language of life, namely the principles underlying protein sequences better than any features suggested by textbooks and prediction methods. The exception is evolutionary information, however, that information is not available on the level of a single sequence.