Maron, J

Abstract Rationale Pulmonary hypertension (PH) is classified into five groups. Diagnosis requires multiple tests delaying treatment. This study compared patient and diagnostic characteristics for pulmonary arterial hypertension (PAH; Group 1 PH), Group 2-5 PH, and non-PH using machine-learning in an Electronic Health Record (EHR) dataset. This is novel as prior studies use claims data which may lack hemodynamic parameters. Methods This study analyzed adult patient records from a HIPPA-complaint EHR dataset from Mayo Clinic (2015-2024). Cohort definitions: PAH: right heart catheterization (RHC) data [mPAP>20 mmHg, PVR>2 WU, PCWP≤15 mmHg]; ≥2 PAH ICD diagnosis codes or positive natural language processing (NLP) sentiments separated by ≥ 30 days; and use of at least one PAH medication within six months after index date. Group 2-5 PH: RHC (mPAP>20mmHg) and ICD codes or positive NLP sentiment separated by ≥ 30 days for corresponding PH subtype. Non-PH: mPAP ≤ 20 mmHg and/or TRV ≤ 2.8 m/s. Index was the earliest NLP or ICD for PAH, earliest RHC for group 2-5 PH and most recent TRV/ mPAP for the non-PH cohort. Clinical variables were analyzed one-year before study index acknowledging data missingness as a limitation of retrospective dataset studies. Results 593 PAH, 817 Group 2-5 PH (67% with group 2 PH), and 11,600 non-PH patients (1:10 propensity-matched) were identified. Age, race, and symptoms were similar across cohorts. PAH had the highest proportion of females (60%) [Group 2-5 PH (50%) and non-PH cohort (53%)]. Clinical characteristic comparison is presented in Table 1. Comorbidities differed across cohorts. RHC: As expected, compared to Group 2-5, the PAH cohort had higher mPAP and PVR and lower PCWP. confirming a distinctly pre-capillary profile. ECHO: Compared to Group 2-5 PH, the PAH cohort had more pronounced right ventricular (RV) pressure overload and dysfunction and more impaired RV longitudinal strain, while TAPSE was similar. At the same time, left heart structure was less affected in PAH compared to Group 2-5. Among laboratory tests, serum creatinine was higher in Group 2-5 and ANA was higher in PAH cohort. Compared to the no-PH cohort, PAH and Group 2-5 cohorts exhibited clinically non-normal RHC and ECHO findings. Conclusion Although symptoms and demographic characteristics were similar among patients with and without PAH, unique RHC, ECHO, laboratory profiles and comorbidities differentiated PAH patients. Screening techniques using these distinct profiles could be considered to facilitate earlier diagnosis and treatment of PAH in clinical settings. This abstract is funded by: Merck & Co., Inc., Rahway, NJ, USA

Abstract Rationale Pulmonary arterial hypertension (PAH) is a rare and progressive disease, where the guideline-recommended diagnostic confirmation is via right heart catheterization (RHC). However, due to non-specific symptomology, patients often experience delays in diagnosis that potentially impact disease outcomes. We sought to develop a machine learning (ML) based approach for early and appropriate detection of PAH. Methods Three cohorts were derived from Mayo Clinic electronic health record (EHR): confirmed PAH (N = 1,169), No PH (N = 11,600) and Other PH (Group 2-5 PH, N = 817), based on 2022 ESC/ERS guidelines. The study index was defined as the earliest diagnosis dates for PH cohorts (PAH & Other PH) or the latest PH negative echocardiogram (ECHO) or RHC for the No-PH cohort. Patients included in the study had at least one symptom (dyspnea, fatigue, syncope, chest pain, or lower extremity swelling) within one year prior to the study index. A two-step sequential ML modelling strategy was applied: Step 1 differentiated PH (PAH + Other PH) from No PH and step 2 differentiated PAH from Other PH. Models were trained using a combination of structured data, unstructured clinical notes to detect symptoms using natural language processing and ECG waveforms captured one year prior to study index. Models were tested both with and without ECHO parameters. SHAP (SHapley Additive explanations) values were used to determine important features contributing to models (without ECG waveforms). Results Models including ECHO parameters performed better compared to those without (Figure 1). The top feature impacting model without ECHO parameters was dyspnea in the preceding three months followed by heart failure and systolic blood pressure for the PH vs no PH; and T wave peak in V1 and ANA level in PAH vs other PH cohort. The top feature impacting model with ECHO was tricuspid regurgitation velocity followed by heart failure and right atrial pressure for the PH vs no PH; and mean blood pressure and left ventricular diameter in the PAH vs other PH cohort. Figure 1: Model performance in the two-step sequential approach PR-AUC=Precision Recall-Area Under Curve Conclusion This study demonstrates that ML models can aid in the identification of PAH in symptomatic patient populations. While ECG can identify patients for further testing, ECHO can further increase the accuracy and the need to refer patients to RHC for confirming diagnosis of PAH. This abstract is funded by: Merck & Co., Inc., Rahway, NJ, USA