Predictive modeling of lean body mass, appendicular lean mass, and appendicular skeletal muscle mass using machine learning techniques: A comprehensive analysis utilizing NHANES data and the Look AHEAD study.

This study addresses the pressing need for improved methods to predict lean mass in adults, and in particular lean body mass (LBM), appendicular lean mass (ALM), and appendicular skeletal muscle mass (ASMM) for the early detection and management of sarcopenia, a condition characterized by muscle loss and dysfunction. Sarcopenia presents significant health risks, especially in populations with chronic diseases like cancer and the elderly. Current assessment methods, primarily relying on Dual-energy X-ray absorptiometry (DXA) scans, lack widespread applicability, hindering timely intervention. Leveraging machine learning techniques, this research aimed to develop and validate predictive models using data from the National Health and Nutrition Examination Survey (NHANES) and the Action for Health in Diabetes (Look AHEAD) study. The models were trained on anthropometric data, demographic factors, and DXA-derived metrics to accurately estimate LBM, ALM, and ASMM normalized to weight. Results demonstrated consistent performance across various machine learning algorithms, with LassoNet, a non-linear extension of the popular LASSO method, exhibiting superior predictive accuracy. Notably, the integration of bone mineral density measurements into the models had minimal impact on predictive accuracy, suggesting potential alternatives to DXA scans for lean mass assessment in the general population. Despite the robustness of the models, limitations include the absence of outcome measures and cohorts highly vulnerable to muscle mass loss. Nonetheless, these findings hold promise for revolutionizing lean mass assessment paradigms, offering implications for chronic disease management and personalized health interventions. Future research endeavors should focus on validating these models in diverse populations and addressing clinical complexities to enhance prediction accuracy and clinical utility in managing sarcopenia.

Obesity prediction: Novel machine learning insights into waist circumference accuracy.

AIMS: This study aims to enhance the precision of obesity risk assessments by improving the accuracy of waist circumference predictions using machine learning techniques. METHODS: We utilized data from the NHANES and Look AHEAD studies, applying machine learning algorithms augmented with uncertainty quantification. Our approach centered on conformal prediction techniques, which provide a methodological basis for generating prediction intervals that reflect uncertainty levels. This method allows for constructing intervals expected to contain the true waist circumference values with a high degree of probability. RESULTS: The application of conformal predictions yielded high coverage rates, achieving 0.955 for men and 0.954 for women in the NHANES dataset. These rates surpassed the expected performance benchmarks and demonstrated robustness when applied to the Look AHEAD dataset, maintaining coverage rates of 0.951 for men and 0.952 for women. Traditional point prediction models did not show such high consistency or reliability. CONCLUSIONS: The findings support the integration of waist circumference into standard clinical practice for obesity-related risk assessments using machine learning approaches.

Evaluation and mitigation of cognitive biases in medical language models.

Increasing interest in applying large language models (LLMs) to medicine is due in part to their impressive performance on medical exam questions. However, these exams do not capture the complexity of real patient-doctor interactions because of factors like patient compliance, experience, and cognitive bias. We hypothesized that LLMs would produce less accurate responses when faced with clinically biased questions as compared to unbiased ones. To test this, we developed the BiasMedQA dataset, which consists of 1273 USMLE questions modified to replicate common clinically relevant cognitive biases. We assessed six LLMs on BiasMedQA and found that GPT-4 stood out for its resilience to bias, in contrast to Llama 2 70B-chat and PMC Llama 13B, which showed large drops in performance. Additionally, we introduced three bias mitigation strategies, which improved but did not fully restore accuracy. Our findings highlight the need to improve LLMs' robustness to cognitive biases, in order to achieve more reliable applications of LLMs in healthcare.

DeepAction: a MATLAB toolbox for automated classification of animal behavior in video.

The identification of animal behavior in video is a critical but time-consuming task in many areas of research. Here, we introduce DeepAction, a deep learning-based toolbox for automatically annotating animal behavior in video. Our approach uses features extracted from raw video frames by a pretrained convolutional neural network to train a recurrent neural network classifier. We evaluate the classifier on two benchmark rodent datasets and one octopus dataset. We show that it achieves high accuracy, requires little training data, and surpasses both human agreement and most comparable existing methods. We also create a confidence score for classifier output, and show that our method provides an accurate estimate of classifier performance and reduces the time required by human annotators to review and correct automatically-produced annotations. We release our system and accompanying annotation interface as an open-source MATLAB toolbox.

Patterns of Oncogene Coexpression at Single-Cell Resolution Influence Survival in Lymphoma.

Cancers often overexpress multiple clinically relevant oncogenes, but it is not known if combinations of oncogenes in cellular subpopulations within a cancer influence clinical outcomes. Using quantitative multispectral imaging of the prognostically relevant oncogenes MYC, BCL2, and BCL6 in diffuse large B-cell lymphoma (DLBCL), we show that the percentage of cells with a unique combination MYC+BCL2+BCL6- (M+2+6-) consistently predicts survival across four independent cohorts (n = 449), an effect not observed with other combinations including M+2+6+. We show that the M+2+6- percentage can be mathematically derived from quantitative measurements of the individual oncogenes and correlates with survival in IHC (n = 316) and gene expression (n = 2,521) datasets. Comparative bulk/single-cell transcriptomic analyses of DLBCL samples and MYC/BCL2/BCL6-transformed primary B cells identify molecular features, including cyclin D2 and PI3K/AKT as candidate regulators of M+2+6- unfavorable biology. Similar analyses evaluating oncogenic combinations at single-cell resolution in other cancers may facilitate an understanding of cancer evolution and therapy resistance. SIGNIFICANCE: Using single-cell-resolved multiplexed imaging, we show that selected subpopulations of cells expressing specific combinations of oncogenes influence clinical outcomes in lymphoma. We describe a probabilistic metric for the estimation of cellular oncogenic coexpression from IHC or bulk transcriptomes, with possible implications for prognostication and therapeutic target discovery in cancer. This article is highlighted in the In This Issue feature, p. 1027.

Unique features of stimulus-based probabilistic reversal learning.

Reversal learning paradigms are widely used assays of behavioral flexibility with their probabilistic versions being more amenable to studying integration of reward outcomes over time. Prior research suggests differences between initial and reversal learning, including higher learning rates, a greater need for inhibitory control, and more perseveration after reversals. However, it is not well-understood what aspects of stimulus-based reversal learning are unique to reversals, and whether and how observed differences depend on reward probability. Here, we used a visual probabilistic discrimination and reversal learning paradigm where male and female rats selected between a pair of stimuli associated with different reward probabilities. We compared accuracy, rewards collected, omissions, latencies, win-stay/lose-shift strategies, and indices of perseveration across two different reward probability schedules. We found that discrimination and reversal learning are behaviorally more unique than similar: Fit of choice behavior using reinforcement learning models revealed a lower sensitivity to the difference in subjective reward values (greater exploration) and higher learning rates for the reversal phase. We also found latencies to choose the better option were greater in females than males, but only for the reversal phase. Further, animals employed more win-stay strategies during early discrimination and increased perseveration during early reversal learning. Interestingly, a consistent reward probability group difference emerged with a richer environment associated with longer reward collection latencies than a leaner environment. Future studies should systematically compare the neural correlates of fine-grained behavioral measures to reveal possible dissociations in how the circuitry is recruited in each phase. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

A Case Study of an Effective and Sustainable Antiretroviral Therapy Program in Rural South Africa.

The delivery of high-quality HIV care in rural settings is a global challenge. Despite the successful expansion of antiretroviral therapy (ART) in Africa, viral load (VL) monitoring and ART adherence are poor, especially in rural communities. This article describes a case study of an ART program in the deeply rural Eastern Cape of South Africa. The Zithulele ART Program initiated five innovations over time: (1) establishing district hospital as the logistical hub for all ART care in a rural district, (2) primary care clinic delivery of prepackaged ART and chronic medications for people living with HIV (PLH), (3) establishing central record keeping, (4) incentivizing VL monitoring, and (5) providing hospital-based outpatient care for complex cases. Using a pharmacy database, on-time VL monitoring and viral suppression were evaluated for 882 PLH initiating ART in the Zithulele catchment area in 2013. Among PLH initiating ART, 12.5% (n = 110) were lost to follow-up, 7.7% (n = 68) transferred out of the region, 10.2% (n = 90) left the program and came back at a later date, and 4.0% (n = 35) died. Of the on-treatment population, 82.9% (n = 480/579) had VL testing within 7 months and 92.6% (n = 536/579) by 1 year. Viral suppression was achieved in 85.2% of those tested (n = 457/536), or 78.9% (n = 457/579) overall. The program's VL testing and suppression rates appear about twice as high as national data and data from other rural centers in South Africa, despite fewer resources than other programs. Simple system innovations can ensure high rates of VL testing and suppression, even in rural health facilities.