Physiology as a Lingua Franca for Clinical Machine Learning

Sarma, Gopal P.; Reinertsen, Erik; Aguirre, Aaron D.; Anderson, Chris; Batra, Puneet; Choi, Seung-Hoan; Achille, Paolo Di; Diamant, Nathaniel; Ellinor, Patrick T.; Emdin, Connor A.; Fahed, Akl C.; Friedman, Samuel; Harrington, Lia X.; Ho, Jennifer E.; Khera, Amit V.; Khurshid, Shaan; Klarqvist, Marcus D. R.; Lubitz, Steve A.; Philippakis, Anthony; Reeder, Christopher; Stultz, Collin M.; Westover, Brandon

doi:10.1016/j.patter.2020.100017

Cited by 9 publications

(8 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The 2D CNNs were optimized with backpropagation and Adaptive Moment stochastic gradient descent (ADAM). The models were implemented in tensorflow version 2.1 using the ML4CVD modeling framework (Sarma et al, 2020). The python package hyperopt was used for Bayesian hyperparameter optimization of the model architecture to select the width, depth, activation function, and the size of each residual block in the CNN.…”

Section: Informed Consent and Study Approvalmentioning

confidence: 99%

Machine learning enables new insights into clinical significance of and genetic contributions to liver fat accumulation

Haas

Friedman

Emdin

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Excess accumulation of liver fat - termed hepatic steatosis when fat accounts for > 5.5% of liver content - is a leading risk factor for end-stage liver disease and is strongly associated with important cardiometabolic disorders. Using a truth dataset of 4,511 UK Biobank participants with liver fat previously quantified via abdominal MRI imaging, we developed a machine learning algorithm to quantify liver fat with correlation coefficients of 0.97 and 0.99 in hold-out testing datasets and applied this algorithm to raw imaging data from an additional 32,192 participants. Among all 36,703 individuals with abdominal MRI imaging, median liver fat was 2.2%, with 6,250 (17%) meeting criteria for hepatic steatosis. Although individuals afflicted with hepatic steatosis were more likely to have been diagnosed with conditions such as obesity or diabetes, a prediction model based on clinical data alone without imaging could not reliably estimate liver fat content. To identify genetic drivers of variation in liver fat, we first conducted a common variant association study of 9.8 million variants, confirming three known associations for variants in the TM6SF2, APOE, and PNPLA3 genes and identifying five new variants associated with increased hepatic fat in or near the MARC1, ADH1B, TRIB1, GPAM and MAST3 genes. A polygenic score that integrated information from each of these eight variants was strongly associated with future clinical diagnosis of liver diseases. Next, we performed a rare variant association study in a subset of 11,021 participants with gene sequencing data available, identifying an association between inactivating variants in the APOB gene and substantially lower LDL cholesterol, but more than 10-fold increased risk of steatosis. Taken together, these results provide proof of principle for the use of machine learning algorithms on raw imaging data to enable epidemiologic studies and genetic discovery.

show abstract

Section: Informed Consent and Study Approvalmentioning

confidence: 99%

Machine learning enables new insights into clinical significance of and genetic contributions to liver fat accumulation

Haas

Friedman

Emdin

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…By expanding physiological knowledge of the behavior of the right ventricle under the progression of PAH, this model provides insights into the causal mechanisms of increased oxygen extraction and how high isovolumic power reduces energy available for stroke power, thereby reducing cardiac output. Evidence of the importance of improving physiological domain knowledge for development of statistical or machine learning clinical risk models is well‐documented (Roe et al, 2020; Sarma et al, 2020).…”

Section: Discussionmentioning

confidence: 99%

Right ventricular myocardial energetic model for evaluating right heart function in pulmonary arterial hypertension

Scott

Tembulkar

Lee

et al. 2022

Physiological Reports

View full text Add to dashboard Cite

Background: Pulmonary arterial hypertension (PAH) increases right ventricular (RV) workload and decreases myocardial oxygen reserve, eventually leading to poor cardiac output. This study created and assessed a novel model of RV work output based on RV hemodynamics and oxygen supply, allowing new insight into causal mechanisms of RV dysfunction. Methods: The RV function model was built upon an earlier, left ventricular model and further adjusted for more accurate clinical use. The model assumes that RV total power output (1) is the sum of isovolumic and stroke power and(2) is linearly related to its right coronary artery oxygen supply. Thus, when right coronary artery flow is limited or isovolumic power is elevated, less energy is available for producing cardiac output. The original and adjusted models were validated via data from patients with idiopathic PAH (n = 14) and large animals (n = 6) that underwent acute pulmonary banding with or without hypoxia.Results: Both models demonstrated strong, significant correlations between RV oxygen consumption rate and RV total power output for PAH patients (original model, R 2 = 0.66; adjusted model, R 2 = 0.78) and sheep (original, R 2 = 0.85; adjusted, R 2 = 0.86). Furthermore, the models demonstrate a significant inverse relationship between required oxygen consumption and RV efficiency (stroke power/total power) (p < 0.001). Lastly, higher NYHA class was indicative of lower RV efficiency and higher oxygen consumption (p = 0.013). Conclusion:Right ventricular total power output can be accurately estimated directly from pulmonary hemodynamics and right coronary perfusion during PAH. This model highlights the increased vulnerability of PAH patients with compromised right coronary flow coupled with high afterload.

show abstract

“…Dynamic Time Warping (DTW) quantified inter‐trajectory distances, using an implementation from the fastdtw python package (Salvador & Chan, 2007). The Broad Institute's ML4H tools were used for model evaluation and tensor‐mapping (Friedman et al., 2020; Sarma et al., 2020).…”

Section: Methodsmentioning

confidence: 99%

Trajectories of sentiment in 11,816 psychoactive narratives

Friedman,

Ballentine

2023

Human Psychopharmacology

View full text Add to dashboard Cite

ObjectiveCan machine learning (ML) enable data‐driven discovery of how changes in sentiment correlate with different psychoactive experiences? We investigate by training models directly on text testimonials from a diverse 52‐drug pharmacopeia.MethodsUsing large language models (i.e. BERT) and 11,816 publicly‐available testimonials, we predicted 28‐dimensions of sentiment across each narrative, and then validated these predictions with adjudication by a clinical psychiatrist. BERT was then fine‐tuned to predict biochemical and demographic information from these narratives. Lastly, canonical correlation analysis linked the drugs' receptor affinities with word usage, revealing 11 statistically‐significant latent receptor‐experience factors, each mapped to a 3D cortical Atlas.ResultsThese methods elucidate a neurobiologically‐informed, sequence‐sensitive portrait of drug‐induced subjective experiences. The models' results converged, revealing a pervasive distinction between the universal psychedelic heights of feeling in contrast to the grim, mundane, and personal experiences of addiction and mental illness. Notably, MDMA was linked to “Love”, DMT and 5‐MeO‐DMT to “Mystical Experiences” and “Entities and Beings”, and other tryptamines to “Surprise”, “Curiosity” and “Realization".ConclusionsML methods can create unified and robust quantifications of subjective experiences with many different psychoactive substances and timescales. The representations learned are evocative and mutually confirmatory, indicating great potential for ML in characterizing psychoactivity.

show abstract

Physiology as a Lingua Franca for Clinical Machine Learning

Cited by 9 publications

References 6 publications

Machine learning enables new insights into clinical significance of and genetic contributions to liver fat accumulation

Machine learning enables new insights into clinical significance of and genetic contributions to liver fat accumulation

Right ventricular myocardial energetic model for evaluating right heart function in pulmonary arterial hypertension

Trajectories of sentiment in 11,816 psychoactive narratives

Contact Info

Product

Resources

About