Revolutionizing Cardiology with Words: Unveiling the Impact of Large Language Models in Medical Science Writing.

Large language models (LLMs) are a unique form of machine learning that facilitates inputs of unstructured text/numerical information for meaningful interpretation and prediction. Recently, LLMs have become commercialized, allowing the average person to access these incredibly powerful tools. Early adopters focused on LLM use in performing logical tasks, including but not limited to generating titles, identifying keywords, summarizing text, initial editing of scientific work, improving statistical protocols, and performing statistical analysis. More recently, LLMs have been expanded to clinical practice and academia to perform higher cognitive and creative tasks. LLMs provide personalized assistance in learning, facilitate the management of electronic medical records, and offer valuable insights into clinical decision-making in cardiology. They enhance patient education by explaining intricate medical conditions in lay terms, have a vast library of knowledge to help clinicians expedite administrative tasks, provide useful feedback regarding content of scientific writing, and assist in the peer review process. Despite their impressive capabilities, LLMs are not without limitations. They are susceptible to generating incorrect or plagiarized content, face challenges in handling tasks without detailed prompts and lack originality. These limitations underscore the importance of human oversight in utilizing LLMs in medical science and clinical practice. As LLMs continue to evolve, addressing these challenges will be crucial in maximizing their potential benefits while mitigating risks. This review explores the functions, opportunities, and constraints of LLMs, with a focus on their impact on cardiology, illustrating both the transformative power and the boundaries of current technology in medicine.

A novel multi-task machine learning classifier for rare disease patterning using cardiac strain imaging data.

To provide accurate predictions, current machine learning-based solutions require large, manually labeled training datasets. We implement persistent homology (PH), a topological tool for studying the pattern of data, to analyze echocardiography-based strain data and differentiate between rare diseases like constrictive pericarditis (CP) and restrictive cardiomyopathy (RCM). Patient population (retrospectively registered) included those presenting with heart failure due to CP (n = 51), RCM (n = 47), and patients without heart failure symptoms (n = 53). Longitudinal, radial, and circumferential strains/strain rates for left ventricular segments were processed into topological feature vectors using Machine learning PH workflow. In differentiating CP and RCM, the PH workflow model had a ROC AUC of 0.94 (Sensitivity = 92%, Specificity = 81%), compared with the GLS model AUC of 0.69 (Sensitivity = 65%, Specificity = 66%). In differentiating between all three conditions, the PH workflow model had an AUC of 0.83 (Sensitivity = 68%, Specificity = 84%), compared with the GLS model AUC of 0.68 (Sensitivity = 52% and Specificity = 76%). By employing persistent homology to differentiate the "pattern" of cardiac deformations, our machine-learning approach provides reasonable accuracy when evaluating small datasets and aids in understanding and visualizing patterns of cardiac imaging data in clinically challenging disease states.

Artificial intelligence augmented home sleep apnea testing device study (AISAP study).

STUDY OBJECTIVE: This study aimed to prospectively validate the performance of an artificially augmented home sleep apnea testing device (WVU-device) and its patented technology. METHODOLOGY: The WVU-device, utilizing patent pending (US 20210001122A) technology and an algorithm derived from cardio-pulmonary physiological parameters, comorbidities, and anthropological information was prospectively compared with a commercially available and Center for Medicare and Medicaid Services (CMS) approved home sleep apnea testing (HSAT) device. The WVU-device and the HSAT device were applied on separate hands of the patient during a single night study. The oxygen desaturation index (ODI) obtained from the WVU-device was compared to the respiratory event index (REI) derived from the HSAT device. RESULTS: A total of 78 consecutive patients were included in the prospective study. Of the 78 patients, 38 (48%) were women and 9 (12%) had a Fitzpatrick score of 3 or higher. The ODI obtained from the WVU-device corelated well with the HSAT device, and no significant bias was observed in the Bland-Altman curve. The accuracy for ODI > = 5 and REI > = 5 was 87%, for ODI> = 15 and REI > = 15 was 89% and for ODI> = 30 and REI of > = 30 was 95%. The sensitivity and specificity for these ODI /REI cut-offs were 0.92 and 0.78, 0.91 and 0.86, and 0.94 and 0.95, respectively. CONCLUSION: The WVU-device demonstrated good accuracy in predicting REI when compared to an approved HSAT device, even in patients with darker skin tones.

A Deep Patient-Similarity Learning Framework for the Assessment of Diastolic Dysfunction in Elderly Patients.

BACKGROUND AND AIMS: Age-related changes in cardiac structure and function are well recognized and make the clinical determination of abnormal LV diastolic function (LVDD) particularly challenging in the elderly. We investigated whether a deep neural network (DeepNN) model[70] of LVDD, previously validated in a younger cohort, can be implemented in an older population to predict incident heart failure (HF). METHODS: A previously developed DeepNN was tested on 5,596 older participants (66-90 years; 57% female; 20% black) from the Atherosclerosis Risk in Communities study. The association of DeepNN predictions with HF or all-cause death for the American College of Cardiology Foundation/American Heart Association Stage A/B (n = 4,054) and Stage C/D (n = 1,542) subgroups was assessed. RESULTS: The DeepNN-predicted High-Risk compared to the Low-Risk phenogroup demonstrated an increased incidence of HF and death for both Stage A/B and Stage C/D (log-rank p < 0.0001 for all). In multivariable analyses, the High-Risk phenogroup remained an independent predictor of HF and death in both Stages A/B (adjusted hazard ratio (HR) [95% confidence interval], 6.52[4.20-10.13] and 2.21(1.68-2.91), both p < 0.0001) and Stage C/D (6.51[4.06-10.44] and 1.03(1.00-1.06), both p < 0.0001) respectively. In addition, DeepNN showed incremental value over the 2016 ASE/EACVI guidelines (Net reclassification index, 0.5[CI:0.4-0.6], p < 0.001; C-statistic improvement, DeepNN [0.76] vs. ASE/EACVI [0.70], p < 0.001) overall and maintained across stage-groups. CONCLUSIONS: Despite training with a younger cohort, a deep patient-similarity-based learning framework for assessing LVDD provides a robust prediction of all-cause death and incident HF for older patients.

The Role of Artificial Intelligence in Echocardiography: A Clinical Update.

PURPOSE OF REVIEW: In echocardiography, there has been robust development of artificial intelligence (AI) tools for image recognition, automated measurements, image segmentation, and patient prognostication that has created a monumental shift in the study of AI and machine learning models. However, integrating these measurements into complex disease recognition and therapeutic interventions remains challenging. While the tools have been developed, there is a lack of evidence regarding implementing heterogeneous systems for guiding clinical decision-making and therapeutic action. RECENT FINDINGS: Newer AI modalities have shown concrete positive data in terms of user-guided image acquisition and processing, precise determination of both basic and advanced quantitative echocardiographic features, and the potential to construct predictive models, all with the possibility of seamless integration into clinical decision support systems. AI in echocardiography is a powerful and ever-growing tool with the potential for revolutionary effects on the practice of cardiology. In this review article, we explore the growth of AI and its applications in echocardiography, along with clinical implications and the associated regulatory, legal, and ethical considerations.

A novel breakthrough in wrist-worn transdermal troponin-I-sensor assessment for acute myocardial infarction.

Aims: Clinical differentiation of acute myocardial infarction (MI) from unstable angina and other presentations mimicking acute coronary syndromes (ACS) is critical for implementing time-sensitive interventions and optimizing outcomes. However, the diagnostic steps are dependent on blood draws and laboratory turnaround times. We tested the clinical feasibility of a wrist-worn transdermal infrared spectrophotometric sensor (transdermal-ISS) in clinical practice and assessed the performance of a machine learning algorithm for identifying elevated high-sensitivity cardiac troponin-I (hs-cTnI) levels in patients hospitalized with ACS. Methods and results: We enrolled 238 patients hospitalized with ACS at five sites. The final diagnosis of MI (with or without ST elevation) and unstable angina was adjudicated using electrocardiography (ECG), cardiac troponin (cTn) test, echocardiography (regional wall motion abnormality), or coronary angiography. A transdermal-ISS-derived deep learning model was trained (three sites) and externally validated with hs-cTnI (one site) and echocardiography and angiography (two sites), respectively. The transdermal-ISS model predicted elevated hs-cTnI levels with areas under the receiver operator characteristics of 0.90 [95% confidence interval (CI), 0.84-0.94; sensitivity, 0.86; and specificity, 0.82] and 0.92 (95% CI, 0.80-0.98; sensitivity, 0.94; and specificity, 0.64), for internal and external validation cohorts, respectively. In addition, the model predictions were associated with regional wall motion abnormalities [odds ratio (OR), 3.37; CI, 1.02-11.15; P = 0.046] and significant coronary stenosis (OR, 4.69; CI, 1.27-17.26; P = 0.019). Conclusion: A wrist-worn transdermal-ISS is clinically feasible for rapid, bloodless prediction of elevated hs-cTnI levels in real-world settings. It may have a role in establishing a point-of-care biomarker diagnosis of MI and impact triaging patients with suspected ACS.

Evaluating the Efficacy and Safety of EZC Pak, a 5-Day Combination Echinacea-Zinc-Vitamin C Dose Pack with or without Vitamin D, in the Management of Outpatient Upper Respiratory Infections.

Background: Growing antibiotic resistance is among the most serious threats to public health, with antibiotic misuse considered a leading driver of the problem. One of the largest areas of misuse is in outpatient upper respiratory infections (URIs). The purpose of this research is to evaluate the efficacy of EZC Pak, a combination Echinacea-Zinc-Vitamin C dose pack with or without Vitamin D, on the duration of illness and symptom severity of non-specific URIs as an alternative to antibiotics when none are deemed clinically necessary. A secondary analysis was carried out on patient satisfaction. Methods: A total of 360 patients across the United States were enrolled and randomized in a double-blind manner across two intervention groups, EZC Pak, EZC Pak+Vitamin D, and one placebo group. The study utilized a smartphone-based app to capture data. Once a participant reported the first URI symptom, they were instructed to take the intervention as directed and complete the daily symptom survey score until their symptoms resolved. Results: The average EZC Pak participant recovered 1.39 days (90% CI 1.05 to 1.73) faster than the average placebo participant (p=0.017). The average EZC Pak participant reported a 17.43% (90% CI 17.1 to 17.8) lower symptom severity score versus placebo (p=0.029). EZC Pak users reported 2.9 times higher patient satisfaction versus placebo users (p=0.012). The addition of Vitamin D neither benefited nor harmed illness duration or symptom severity. Conclusion: The findings support the potential use of EZC Pak as an alternative to patient request for antibiotics when none are deemed clinically necessary at the time of initial clinical presentation. The decision to replete vitamin D in the acute phase of URI is an individualized decision left to the patient and their clinician. EZC Pak may play a critical role in improving outpatient URI management and antibiotic stewardship (ClinicalTrials.gov number, NCT04943575).

Integrating Echocardiography Parameters With Explainable Artificial Intelligence for Data-Driven Clustering of Primary Mitral Regurgitation Phenotypes.

BACKGROUND: Primary mitral regurgitation (MR) is a heterogeneous clinical disease requiring integration of echocardiographic parameters using guideline-driven recommendations to identify severe disease. OBJECTIVES: The purpose of this preliminary study was to explore novel data-driven approaches to delineate phenotypes of MR severity that benefit from surgery. METHODS: The authors used unsupervised and supervised machine learning and explainable artificial intelligence (AI) to integrate 24 echocardiographic parameters in 400 primary MR subjects from France (n = 243; development cohort) and Canada (n = 157; validation cohort) followed up during a median time of 3.2 years (IQR: 1.3-5.3 years) and 6.8 (IQR: 4.0-8.5 years), respectively. The authors compared the phenogroups' incremental prognostic value over conventional MR profiles and for the primary endpoint of all-cause mortality incorporating time-to-mitral valve repair/replacement surgery as a covariate for survival analysis (time-dependent exposure). RESULTS: High-severity (HS) phenogroups from the French cohort (HS: n = 117; low-severity [LS]: n = 126) and the Canadian cohort (HS: n = 87; LS: n = 70) showed improved event-free survival in surgical HS subjects over nonsurgical subjects (P = 0.047 and P = 0.020, respectively). A similar benefit of surgery was not seen in the LS phenogroup in both cohorts (P = 0.70 and P = 0.50, respectively). Phenogrouping showed incremental prognostic value in conventionally severe or moderate-severe MR subjects (Harrell C statistic improvement; P = 0.480; and categorical net reclassification improvement; P = 0.002). Explainable AI specified how each echocardiographic parameter contributed to phenogroup distribution. CONCLUSIONS: Novel data-driven phenogrouping and explainable AI aided in improved integration of echocardiographic data to identify patients with primary MR and improved event-free survival after mitral valve repair/replacement surgery.

Quantitative single photon emission computed tomography derived standardized uptake values on 99mTc-PYP scan in patients with suspected ATTR cardiac amyloidosis.

Technetium-99 pyrophosphate scintigraphy (99mTc-PYP) provides qualitative and semiquantitative diagnosis of ATTR cardiac amyloidosis (ATTR-CA) using the Perugini scoring system and heart/contralateral heart ratio (H/CL) on planar imaging. Standardized uptake values (SUV) with quantitative single photon emission computed tomography (xSPECT/CT) can offer superior diagnostic accuracy and quantification through precise myocardial contouring that enhances assessment of ATTR-CA burden. We examined the correlation of xSPECT/CT SUVs with Perugini score and H/CL ratio. We also assessed SUV correlation with cardiac magnetic resonance (CMR), echocardiographic, and baseline clinical characteristics. Retrospective review of 78 patients with suspected ATTR-CA that underwent 99mTc-PYP scintigraphy with xSPECT/CT. Patients were grouped off Perugini score (Grade 0-1 and Grade 2-3), H/CL ratio (≥ 1.5 and < 1.5). Two cohorts were also created: myocardium SUVmax > 1.88 and ≤ 1.88 at 1-hour based off an AUC curve with 1.88 showing the greatest sensitivity and specificity. Cardiac SUV retention index was calculated as [SUVmax myocardium/SUVmax vertebrae] × SUVmax paraspinal muscle. Primary outcome was myocardium SUVmax at 1-hour correlation with Perugini grades, H/CL ratio, CMR, and echocardiographic data. Higher Perugini Grades corresponded with higher myocardium SUVmax values, especially when comparing Perugini Grade 3 to Grade 2 and 1 (3.03 ± 2.1 vs 0.59 ± 0.97 and 0.09 ± 0.2, P < 0.001). Additionally, patients with H/CL ≥ 1.5 had significantly higher myocardium SUVmax compared to patients with H/CL ≤ 1.5 (2.92 ± 2.18 vs 0.35 ± 0.60, P < 0.01). Myocardium SUVmax at 1-hour strongly correlated with ECV (r = 0.91, P = 0.001), pre-contrast T1 map values (r = 0.66, P = 0.037), and left ventricle mass index (r = 0.80, P = 0.002) on CMR. SUVs derived from 99mTc-PYP scintigraphy with xSPECT/CT provides a discriminatory and quantitative method to diagnose and assess ATTR-CA burden. These findings strongly correlate with CMR.

Ultrasonic Texture Features for Assessing Cardiac Remodeling and Dysfunction.

BACKGROUND: Changes in cardiac size, myocardial mass, cardiomyocyte appearance, and, ultimately, the function of the entire organ are interrelated features of cardiac remodeling that profoundly affect patient outcomes. OBJECTIVES: This study proposes that the application of radiomics for extracting cardiac ultrasonic textural features (ultrasomics) can aid rapid, automated assessment of left ventricular (LV) structure and function without requiring manual measurements. METHODS: This study developed machine-learning models using cardiac ultrasound images from 1,915 subjects in 3 clinical cohorts: 1) an expert-annotated cardiac point-of-care-ultrasound (POCUS) registry (n = 943, 80% training/testing and 20% internal validation); 2) a prospective POCUS cohort for external validation (n = 275); and 3) a prospective external validation on high-end ultrasound systems (n = 484). In a type 2 diabetes murine model, echocardiography of wild-type (n = 10) and Leptr-/- (n = 8) mice were assessed longitudinally at 3 and 25 weeks, and ultrasomics features were correlated with histopathological features of hypertrophy. RESULTS: The ultrasomics model predicted LV remodeling in the POCUS and high-end ultrasound external validation studies (area under the curve: 0.78 [95% CI: 0.68-0.88] and 0.79 [95% CI: 0.73-0.86], respectively). Similarly, the ultrasomics model predicted LV remodeling was significantly associated with major adverse cardiovascular events in both cohorts (P < 0.0001 and P = 0.0008, respectively). Moreover, on multivariate analysis, the ultrasomics probability score was an independent echocardiographic predictor of major adverse cardiovascular events in the high-end ultrasound cohort (HR: 8.53; 95% CI: 4.75-32.1; P = 0.0003). In the murine model, cardiomyocyte hypertrophy positively correlated with 2 ultrasomics biomarkers (R2 = 0.57 and 0.52, Q < 0.05). CONCLUSIONS: Cardiac ultrasomics-based biomarkers may aid development of machine-learning models that provide an expert-level assessment of LV structure and function.

Electrocardiogram-Based Machine Learning Emulator Model for Predicting Novel Echocardiography-Derived Phenogroups for Cardiac Risk-Stratification: A Prospective Multicenter Cohort Study.

Purpose: Electrocardiography (ECG)-derived machine learning models can predict echocardiography (echo)-derived indices of systolic or diastolic function. However, systolic and diastolic dysfunction frequently coexists, which necessitates an integrated assessment for optimal risk-stratification. We explored an ECG-derived model that emulates an echo-derived model that combines multiple parameters for identifying patient phenogroups at risk for major adverse cardiac events (MACE). Methods: In this substudy of a prospective, multicenter study, patients from 3 institutions (n=727) formed an internal cohort, and the fourth institution was reserved as an external test set (n=518). A previously validated patient similarity analysis model was used for labeling the patients as low-/high-risk phenogroups. These labels were utilized for training an ECG-derived deep neural network model to predict MACE risk per phenogroup. After 5-fold cross-validation training, the model was tested on the reserved external dataset. Results: Our ECG-derived model showed robust classification of patients, with area under the receiver operating characteristic curve of 0.86 (95% CI: 0.79-0.91) and 0.84 (95% CI: 0.80-0.87), sensitivity of 80% and 76%, and specificity of 88% and 75% for the internal and external test sets, respectively. The ECG-derived model demonstrated an increased probability for MACE in high-risk vs low-risk patients (21% vs 3%; P<0.001), which was similar to the echo-trained model (21% vs 5%; P<0.001), suggesting comparable utility. Conclusions: This novel ECG-derived machine learning model provides a cost-effective strategy for predicting patient subgroups in whom an integrated milieu of systolic and diastolic dysfunction is associated with a high risk of MACE.

Histopathology of the broad class of carbon nanotubes and nanofibers used or produced in U.S. facilities in a murine model.

BACKGROUND: Multi-walled carbon nanotubes and nanofibers (CNT/F) have been previously investigated for their potential toxicities; however, comparative studies of the broad material class are lacking, especially those with a larger diameter. Additionally, computational modeling correlating physicochemical characteristics and toxicity outcomes have been infrequently employed, and it is unclear if all CNT/F confer similar toxicity, including histopathology changes such as pulmonary fibrosis. Male C57BL/6 mice were exposed to 40 µg of one of nine CNT/F (MW #1-7 and CNF #1-2) commonly found in exposure assessment studies of U.S. facilities with diameters ranging from 6 to 150 nm. Human fibroblasts (0-20 µg/ml) were used to assess the predictive value of in vitro to in vivo modeling systems. RESULTS: All materials induced histopathology changes, although the types and magnitude of the changes varied. In general, the larger diameter MWs (MW #5-7, including Mitsui-7) and CNF #1 induced greater histopathology changes compared to MW #1 and #3 while MW #4 and CNF #2 were intermediate in effect. Differences in individual alveolar or bronchiolar outcomes and severity correlated with physical dimensions and how the materials agglomerated. Human fibroblast monocultures were found to be insufficient to fully replicate in vivo fibrosis outcomes suggesting in vitro predictive potential depends upon more advanced cell culture in vitro models. Pleural penetrations were observed more consistently in CNT/F with larger lengths and diameters. CONCLUSION: Physicochemical characteristics, notably nominal CNT/F dimension and agglomerate size, predicted histopathologic changes and enabled grouping of materials by their toxicity profiles. Particles of greater nominal tube length were generally associated with increased severity of histopathology outcomes. Larger particle lengths and agglomerates were associated with more severe bronchi/bronchiolar outcomes. Spherical agglomerated particles of smaller nominal tube dimension were linked to granulomatous inflammation while a mixture of smaller and larger dimensional CNT/F resulted in more severe alveolar injury.

Deep neural survival networks for cardiovascular risk prediction: The Multi-Ethnic Study of Atherosclerosis (MESA).

BACKGROUND: There is growing interest in utilizing machine learning techniques for routine atherosclerotic cardiovascular disease (ASCVD) risk prediction. We investigated whether novel deep learning survival models can augment ASCVD risk prediction over existing statistical and machine learning approaches. METHODS: 6814 participants from the Multi-Ethnic Study of Atherosclerosis (MESA) were followed over 16 years to assess incidence of all-cause mortality (mortality) or a composite of major adverse events (MAE). Features were evaluated within the categories of traditional risk factors, inflammatory biomarkers, and imaging markers. Data was split into an internal training/testing (four centers) and external validation (two centers). Both machine learning (COXPH, RSF, and lSVM) and deep learning (nMTLR and DeepSurv) models were evaluated. RESULTS: In comparison to the COXPH model, DeepSurv significantly improved ASCVD risk prediction for MAE (AUC: 0.82 vs. 0.80, P ≤ 0.001) and mortality (AUC: 0.87 vs. 0.84, P ≤ 0.001) with traditional risk factors alone. Implementing non-categorical NRI, we noted a >40% increase in correct reclassification compared to the COXPH model for both MAE and mortality (P ≤ 0.05). Assessing the relative risk of participants, DeepSurv was the only learning algorithm to develop a significantly improved risk score criteria, which outcompeted COXPH for both MAE (4.22 vs. 3.61, P = 0.043) and mortality (6.81 vs. 5.52, P = 0.044). The addition of inflammatory or imaging biomarkers to traditional risk factors showed minimal/no significant improvement in model prediction. CONCLUSION: DeepSurv can leverage simple office-based clinical features alone to accurately predict ASCVD risk and cardiovascular outcomes, without the need for additional features, such as inflammatory and imaging biomarkers.

A vital sign-based prediction algorithm for differentiating COVID-19 versus seasonal influenza in hospitalized patients.

Patients with influenza and SARS-CoV2/Coronavirus disease 2019 (COVID-19) infections have a different clinical course and outcomes. We developed and validated a supervised machine learning pipeline to distinguish the two viral infections using the available vital signs and demographic dataset from the first hospital/emergency room encounters of 3883 patients who had confirmed diagnoses of influenza A/B, COVID-19 or negative laboratory test results. The models were able to achieve an area under the receiver operating characteristic curve (ROC AUC) of at least 97% using our multiclass classifier. The predictive models were externally validated on 15,697 encounters in 3125 patients available on TrinetX database that contains patient-level data from different healthcare organizations. The influenza vs COVID-19-positive model had an AUC of 98.8%, and 92.8% on the internal and external test sets, respectively. Our study illustrates the potentials of machine-learning models for accurately distinguishing the two viral infections. The code is made available at https://github.com/ynaveena/COVID-19-vs-Influenza and may have utility as a frontline diagnostic tool to aid healthcare workers in triaging patients once the two viral infections start cocirculating in the communities.

Deep-Learning Models for the Echocardiographic Assessment of Diastolic Dysfunction.

OBJECTIVES: The authors explored a deep neural network (DeepNN) model that integrates multidimensional echocardiographic data to identify distinct patient subgroups with heart failure with preserved ejection fraction (HFpEF). BACKGROUND: The clinical algorithms for phenotyping the severity of diastolic dysfunction in HFpEF remain imprecise. METHODS: The authors developed a DeepNN model to predict high- and low-risk phenogroups in a derivation cohort (n = 1,242). Model performance was first validated in 2 external cohorts to identify elevated left ventricular filling pressure (n = 84) and assess its prognostic value (n = 219) in patients with varying degrees of systolic and diastolic dysfunction. In 3 National Heart, Lung, and Blood Institute-funded HFpEF trials, the clinical significance of the model was further validated by assessing the relationships of the phenogroups with adverse clinical outcomes (TOPCAT [Aldosterone Antagonist Therapy for Adults With Heart Failure and Preserved Systolic Function] trial, n = 518), cardiac biomarkers, and exercise parameters (NEAT-HFpEF [Nitrate's Effect on Activity Tolerance in Heart Failure With Preserved Ejection Fraction] and RELAX-HF [Evaluating the Effectiveness of Sildenafil at Improving Health Outcomes and Exercise Ability in People With Diastolic Heart Failure] pooled cohort, n = 346). RESULTS: The DeepNN model showed higher area under the receiver-operating characteristic curve than 2016 American Society of Echocardiography guideline grades for predicting elevated left ventricular filling pressure (0.88 vs. 0.67; p = 0.01). The high-risk (vs. low-risk) phenogroup showed higher rates of heart failure hospitalization and/or death, even after adjusting for global left ventricular and atrial longitudinal strain (hazard ratio [HR]: 3.96; 95% confidence interval [CI]: 1.24 to 12.67; p = 0.021). Similarly, in the TOPCAT cohort, the high-risk (vs. low-risk) phenogroup showed higher rates of heart failure hospitalization or cardiac death (HR: 1.92; 95% CI: 1.16 to 3.22; p = 0.01) and higher event-free survival with spironolactone therapy (HR: 0.65; 95% CI: 0.46 to 0.90; p = 0.01). In the pooled RELAX-HF/NEAT-HFpEF cohort, the high-risk (vs. low-risk) phenogroup had a higher burden of chronic myocardial injury (p < 0.001), neurohormonal activation (p < 0.001), and lower exercise capacity (p = 0.001). CONCLUSIONS: This publicly available DeepNN classifier can characterize the severity of diastolic dysfunction and identify a specific subgroup of patients with HFpEF who have elevated left ventricular filling pressures, biomarkers of myocardial injury and stress, and adverse events and those who are more likely to respond to spironolactone.

A Machine-Learning Framework to Identify Distinct Phenotypes of Aortic Stenosis Severity.

OBJECTIVES: The authors explored the development and validation of machine-learning models for augmenting the echocardiographic grading of aortic stenosis (AS) severity. BACKGROUND: In AS, symptoms and adverse events develop secondarily to valvular obstruction and left ventricular decompensation. The current echocardiographic grading of AS severity focuses on the valve and is limited by diagnostic uncertainty. METHODS: Using echocardiography (ECHO) measurements (ECHO cohort, n = 1,052), we performed patient similarity analysis to derive high-severity and low-severity phenogroups of AS. We subsequently developed a supervised machine-learning classifier and validated its performance with independent markers of disease severity obtained using computed tomography (CT) (CT cohort, n = 752) and cardiovascular magnetic resonance (CMR) imaging (CMR cohort, n = 160). The classifier's prognostic value was further validated using clinical outcomes (aortic valve replacement [AVR] and death) observed in the ECHO and CMR cohorts. RESULTS: In 1,964 patients from the 3 multi-institutional cohorts, 1,346 (68%) subjects had either nonsevere or discordant AS severity. Machine learning identified 1,117 (57%) patients as having high-severity and 847 (43%) as having low-severity AS. High-severity patients in CT and CMR cohorts had higher valve calcium scores and left ventricular mass and fibrosis, respectively than the low-severity group. In the ECHO cohort, progression to AVR and progression to death in patients who did not receive AVR was faster in the high-severity group. Compared with the conventional classification of disease severity, machine-learning-based severity classification improved discrimination (integrated discrimination improvement: 0.07; 95% confidence interval: 0.02 to 0.12) and reclassification (net reclassification improvement: 0.17; 95% confidence interval: 0.11 to 0.23) for the outcome of AVR at 5 years. For both ECHO and CMR cohorts, we observed prognostic value of the machine-learning classifications for subgroups with asymptomatic, nonsevere or discordant AS. CONCLUSIONS: Machine learning can integrate ECHO measurements to augment the classification of disease severity in most patients with AS, with major potential to optimize the timing of AVR.

Malignant Pleural Mesothelioma Interactome with 364 Novel Protein-Protein Interactions.

Malignant pleural mesothelioma (MPM) is an aggressive cancer affecting the outer lining of the lung, with a median survival of less than one year. We constructed an 'MPM interactome' with over 300 computationally predicted protein-protein interactions (PPIs) and over 2400 known PPIs of 62 literature-curated genes whose activity affects MPM. Known PPIs of the 62 MPM associated genes were derived from Biological General Repository for Interaction Datasets (BioGRID) and Human Protein Reference Database (HPRD). Novel PPIs were predicted by applying the HiPPIP algorithm, which computes features of protein pairs such as cellular localization, molecular function, biological process membership, genomic location of the gene, and gene expression in microarray experiments, and classifies the pairwise features as interacting or non-interacting based on a random forest model. We validated five novel predicted PPIs experimentally. The interactome is significantly enriched with genes differentially ex-pressed in MPM tumors compared with normal pleura and with other thoracic tumors, genes whose high expression has been correlated with unfavorable prognosis in lung cancer, genes differentially expressed on crocidolite exposure, and exosome-derived proteins identified from malignant mesothelioma cell lines. 28 of the interactors of MPM proteins are targets of 147 U.S. Food and Drug Administration (FDA)-approved drugs. By comparing disease-associated versus drug-induced differential expression profiles, we identified five potentially repurposable drugs, namely cabazitaxel, primaquine, pyrimethamine, trimethoprim and gliclazide. Preclinical studies may be con-ducted in vitro to validate these computational results. Interactome analysis of disease-associated genes is a powerful approach with high translational impact. It shows how MPM-associated genes identified by various high throughput studies are functionally linked, leading to clinically translatable results such as repurposed drugs. The PPIs are made available on a webserver with interactive user interface, visualization and advanced search capabilities.

A Vital Sign-based Prediction Algorithm for Differentiating COVID-19 Versus Seasonal Influenza in Hospitalized Patients.

Patients with influenza and SARS-CoV2/Coronavirus disease 2019 (COVID-19) infections have different clinical course and outcomes. We developed and validated a supervised machine learning pipeline to distinguish the two viral infections using the available vital signs and demographic dataset from the first hospital/emergency room encounters of 3,883 patients who had confirmed diagnoses of influenza A/B, COVID-19 or negative laboratory test results. The models were able to achieve an area under the receiver operating characteristic curve (ROC AUC) of at least 97% using our multiclass classifier. The predictive models were externally validated on 15,697 encounters in 3,125 patients available on TrinetX database that contains patient-level data from different healthcare organizations. The influenza vs. COVID-19-positive model had an AUC of 98%, and 92% on the internal and external test sets, respectively. Our study illustrates the potentials of machine-learning models for accurately distinguishing the two viral infections. The code is made available at https://github.com/ynaveena/COVID-19-vs-Influenza and may be have utility as a frontline diagnostic tool to aid healthcare workers in triaging patients once the two viral infections start cocirculating in the communities.