Introduction: Pulmonary hypertension (PH) is defined by mean pulmonary arterial pressure (mPAP) of ≥ 21 mmHg or ≥ 25 mmHg by right heart catheterization (RHC), (2022 and 2015 ERS/ESC guidelines). Based on a recent meta-analysis, the sensitivity and specificity of transthoracic echocardiography (TTE) is 83% and 72% in the 59% of patients where tricuspid regurgitant velocity (TRV) is available. Research Question: Can we develop a point of care method to detect PH, based on cardiac orthogonal voltage gradient (OVG) signals and photoplethysmographic (PPG) signals? Aim: Using OVG and PPG signals from subjects evaluated with RHC or TTE (independent of TRV), we built a machine learned (ML) model that can identify subjects with PH with high accuracy. Methods: 3298 features from PPG and OVG signals were reduced to 216 via univariate testing. ElasticNet and Random Forest models were developed using a coarse grid-search based on 5-fold cross-validation followed by a narrow search when the two algorithms were ensembled, performed on a naive set of subjects with confirmed PH (N=327, mPAP ≥ 21 mmHg; N=252, mPAP ≥ 25 mmHg) vs. subjects with high negative predictive value for absence of PH, i.e., meeting the 2022 ERS/ESC TTE guideline criteria for “Low Probability” of PH adjudicated by a core lab (N=204). The model prediction for the selected hyperparameters was generated for every subject using out-of-fold prediction. Results: Table 1 demonstrates strong performance across gender and PH subgroups. AUC for all PH/Pre-capillary PH based on the 2015 ERS/ESC guidelines is 0.92/0.95, sensitivity 90%/95% and specificity 75%/75%. Using the updated 2022 guidelines, AUC for all PH/Pre-capillary PH is 0.90/0.92, sensitivity 90%/93% and specificity 69%/69%. Conclusion: Clinically relevant performance using a ML algorithm based on non-invasive signals collected at point-of-care is achievable in the identification of PH (and specifically, Pre-capillary PH), not limited by availability of TRV.