Using machine learning to forecast symptom changes among subclinical depression patients receiving stepped care or usual care.

BACKGROUND: Subclinical depression (SD) is a mental health disorder characterized by minor depressive symptoms. Most SD patients are treated in the primary practice, but many respond poorly to treatment at the expense of provider resources. Stepped care approaches are appealing for tiering SD care to efficiently allocate scarce resources while jointly optimizing patient outcomes. However, stepped care can be time inefficient, as some persons may respond poorly and be forced to suffer with their symptoms for prolonged periods. Machine learning can offer insight into optimal treatment paths and inform clinical recommendations for incident patients. METHODS: As part of the Step-Dep trial, participants with SD were randomized to receive stepped care (N=96) or usual care (N=140). Machine learning was used to predict changes in depressive symptoms every three months over a year for each treatment group. RESULTS: Tree-based models were effective in predicting PHQ-9 changes among patients who received stepped care (r=0.35-0.46, MAE=0.14-0.17) and usual care (r=0.34-0.49, MAE=0.15-0.18). Patients who received stepped care were more likely to reduce PHQ-9 scores if they had high PHQ-9 but low HADS-A scores at baseline, a low number of chronic illnesses, and an internal locus of control. LIMITATIONS: Models may suffer from potential overfitting due to sample size limitations. CONCLUSION: Our findings demonstrate the promise of machine learning for predicting changes in depressive symptoms for SD patients receiving different treatments. Trained models can intake incident patient information and predict outcomes to inform personalized care.

Development and Validation of a Model to Identify Critical Brain Injuries Using Natural Language Processing of Text Computed Tomography Reports.

Importance: Clinical text reports from head computed tomography (CT) represent rich, incompletely utilized information regarding acute brain injuries and neurologic outcomes. CT reports are unstructured; thus, extracting information at scale requires automated natural language processing (NLP). However, designing new NLP algorithms for each individual injury category is an unwieldy proposition. An NLP tool that summarizes all injuries in head CT reports would facilitate exploration of large data sets for clinical significance of neuroradiological findings. Objective: To automatically extract acute brain pathological data and their features from head CT reports. Design, Setting, and Participants: This diagnostic study developed a 2-part named entity recognition (NER) NLP model to extract and summarize data on acute brain injuries from head CT reports. The model, termed BrainNERD, extracts and summarizes detailed brain injury information for research applications. Model development included building and comparing 2 NER models using a custom dictionary of terms, including lesion type, location, size, and age, then designing a rule-based decoder using NER outputs to evaluate for the presence or absence of injury subtypes. BrainNERD was evaluated against independent test data sets of manually classified reports, including 2 external validation sets. The model was trained on head CT reports from 1152 patients generated by neuroradiologists at the Yale Acute Brain Injury Biorepository. External validation was conducted using reports from 2 outside institutions. Analyses were conducted from May 2020 to December 2021. Main Outcomes and Measures: Performance of the BrainNERD model was evaluated using precision, recall, and F1 scores based on manually labeled independent test data sets. Results: A total of 1152 patients (mean [SD] age, 67.6 [16.1] years; 586 [52%] men), were included in the training set. NER training using transformer architecture and bidirectional encoder representations from transformers was significantly faster than spaCy. For all metrics, the 10-fold cross-validation performance was 93% to 99%. The final test performance metrics for the NER test data set were 98.82% (95% CI, 98.37%-98.93%) for precision, 98.81% (95% CI, 98.46%-99.06%) for recall, and 98.81% (95% CI, 98.40%-98.94%) for the F score. The expert review comparison metrics were 99.06% (95% CI, 97.89%-99.13%) for precision, 98.10% (95% CI, 97.93%-98.77%) for recall, and 98.57% (95% CI, 97.78%-99.10%) for the F score. The decoder test set metrics were 96.06% (95% CI, 95.01%-97.16%) for precision, 96.42% (95% CI, 94.50%-97.87%) for recall, and 96.18% (95% CI, 95.151%-97.16%) for the F score. Performance in external institution report validation including 1053 head CR reports was greater than 96%. Conclusions and Relevance: These findings suggest that the BrainNERD model accurately extracted acute brain injury terms and their properties from head CT text reports. This freely available new tool could advance clinical research by integrating information in easily gathered head CT reports to expand knowledge of acute brain injury radiographic phenotypes.

Incidence of diabetes mellitus and factors associated with its development in HIV-positive patients over the age of 50.

OBJECTIVE: We sought to determine the incidence and factors associated with development of diabetes mellitus (DM) in older HIV-infected individuals. RESEARCH DESIGN AND METHODS: We analyzed data from people living with HIV (PLWH) ≥50 years of age enrolled in a large urban HIV outpatient clinic in Vancouver, British Columbia. Patients were categorized as having DM if they had random blood sugar ≥11.1 mmol/L, fasting blood sugar ≥7 mmol/L, HbA1C ≥6.5%, antidiabetic medication use during the follow-up period, or medical chart review confirming diagnosis of DM. We estimated the probability of developing DM, adjusting for demographic and clinical factors, using a logistic regression model. RESULTS: Among 1065 PLWH followed for a median of 13 years (25th and 75th percentile (Q1-Q3): 9-18), the incidence of DM was 1.61/100 person-years follow-up. In the analysis of factors associated with new-onset DM (n=703), 88% were male, 38% had a history of injection drug use, 43% were hepatitis C coinfected, and median body mass index was 24 kg/m2 (Q1-Q3: 21-27). Median age at antiretroviral therapy (ART) initiation was 48 years (Q1-Q3: 43-53) and at DM diagnosis was 55 years (Q1-Q3: 50-61). Patients who started ART in 1997-1999 and had a longer exposure to older ART were at the highest risk of developing DM. CONCLUSIONS: Among PLWH aged ≥50 years, the incidence of DM was 1.39 times higher than men in the general Canadian population of similar age. ART initiated in the early years of the epidemic and exposure to older ART appeared to be the main drivers of the development of DM.

Regulation of flagellum number by FliA and FlgM and role in biofilm formation by Rhodobacter sphaeroides.

The FlgM secretion checkpoint plays a crucial role in coordinating bacterial flagellar assembly. Here we identify a new role for FlgM and FliA as part of a complex regulatory network which controls flagellum number and is essential for efficient swimming and biofilm formation in the monotrichous bacterium Rhodobacter sphaeroides.