YC471: PULSE (Predictive Utility For Learning Systems In Evaluating Anemia And Hypertension):A Photoplethysmography (PPG) And Machine Learning–Based Device For Non-Invasive Monitoring Of Hb, HR, SpO2, And BP

Venisse Kyle V. Claudio Tupi National High School

Anemia and hypertension remain prevalent health concerns in low-resource communities with limited diagnostic access. This study developed PULSE, a portable, non-invasive screening device using photoplethysmography and multiple linear regression to estimate hemoglobin, blood pressure, oxygen saturation, and heart rate in real time. PULSE was validated against standard clinical instruments. Results showed no significant differences (p > 0.05) between PULSE and reference devices. The MLR model achieved R² = 0.972, RMSE = 2.18, MAE = 1.60, and MAPE = 1.45% for hemoglobin prediction. Blood pressure estimation attained R² = 0.97 (SBP) and R² = 0.95 (DBP) with mean errors of 2.4 mmHg and 2.7 mmHg. Strong correlations (r = 0.951–0.984) and ICC = 0.981 confirmed high reliability. The system displayed results within 1.82 s (OLED) and 2.14 s (Wi-Fi) with 100% logging accuracy and 93.3% mapping precision. These findings confirm that PULSE is a reliable, accurate, and cost-effective solution for early anemia and hypertension screening.