Wearable Biomedical Final Year Projects 2026
Wearable biomedical systems sense physiology and motion from the body (ECG, PPG, IMU, temperature) and process signals for health insights — often with on-device or edge machine learning. Student projects combine signal processing, feature engineering or deep models, evaluation on public datasets, and optional hardware prototypes (Arduino, ESP32, commercial wearable APIs).
Below: 80+ topics with tools and representative public datasets.
Wearable Biomedical Sensors
Tools & PlatformsBest Wearable Biomedical Project Topics (80+)
Topics with tools and datasets.
| # | Project Topic | Tools | Datasets |
|---|---|---|---|
| ECG Signal Processing & Arrhythmia | |||
| 01 | ECGWearable ECG Preprocessing: Filter, QRS Detection, Beat Segmentation | NeuroKit2 · BioSPPy · Python | MIT-BIH · PhysioNet |
| 02 | ECGArrhythmia Classification from Single-Lead Wearable ECG | scikit-learn · CNN · PyTorch | MIT-BIH Arrhythmia |
| 03 | ECGAtrial Fibrillation Detection from Short Wearable Recordings | Python · RF / LSTM | PhysioNet AF Challenge |
| 04 | ECGHeart Rate Variability (HRV) Feature Pipeline from ECG | NeuroKit2 · hrvanalysis | MIT-BIH · custom ECG |
| 05 | ECGNoise Robustness of Wearable ECG under Motion Artefacts | filter banks · quality indices | Motion-corrupted ECG sets |
| 06 | ECGTransfer Learning from Clinical ECG to Wearable Domains | PyTorch · domain adaptation | MIT-BIH + wearable ECG |
| 07 | ECGReal-Time Beat Classification Demo with Sliding Windows | Python · streaming buffer | MIT-BIH streaming sim |
| 08 | ECGExplainable ECG Classification with Attention / SHAP | PyTorch · SHAP · Grad-CAM | MIT-BIH |
| 09 | ECGMulti-Lead vs Single-Lead Performance Comparison | NeuroKit2 · classical + DL | PTB-XL concepts |
| 10 | ECGECG Quality Assessment for Wearable Acceptance/Rejection | signal quality metrics · ML | Noisy wearable traces |
| PPG, SpO2 & Optical Sensing | |||
| 11 | PPGPPG Peak Detection and Heart Rate Estimation Pipeline | NeuroKit2 · HeartPy · Python | PPG-DaLiA · BIDMC |
| 12 | PPGMotion Artefact Reduction in Wrist PPG during Exercise | adaptive filters · Python | PPG-DaLiA |
| 13 | PPGSpO2 Estimation Concepts from Dual-Wavelength PPG | calibration models · Python | BIDMC · pulse ox sets |
| 14 | PPGPPG-Based Blood Pressure Proxy / Cuffless Concepts | feature regression · ML | PulseDB / cuffless sets |
| 15 | PPGRespiratory Rate Estimation from PPG Waveform | frequency / amplitude modulation | BIDMC · CapnoBase concepts |
| 16 | PPGDeep Learning Heart Rate from Raw PPG Windows | 1D CNN / LSTM · PyTorch | PPG-DaLiA |
| 17 | PPGCross-Device PPG Generalisation (Phone vs Watch Form Factors) | domain shift study | Multi-device PPG logs |
| 18 | PPGSkin Tone and Contact Pressure Effects on PPG Quality | quality metrics · analysis | Multi-subject PPG |
| IMU, Activity Recognition & Gait | |||
| 19 | IMUHuman Activity Recognition (HAR) from Wearable IMU | scikit-learn · CNN · TF | PAMAP2 · UCI HAR |
| 20 | IMUWindowing, Feature Engineering vs End-to-End Deep HAR | Python · comparison study | Opportunity · PAMAP2 |
| 21 | IMUGait Cycle Detection and Step Counting from Waist/Ankle IMU | signal processing · Python | gait IMU datasets |
| 22 | IMUParkinsonian Gait Feature Extraction from Wearable Sensors | kinematic features · ML | Parkinson gait sets |
| 23 | IMUTransition Detection (Sit–Stand–Walk) for Context Awareness | HMM / ML · Python | PAMAP2 transitions |
| 24 | IMUMulti-Sensor Fusion: Phone + Wrist IMU for Robust HAR | fusion models · Python | Opportunity |
| 25 | IMUEnergy Expenditure Proxy from Accelerometry | regression · MET concepts | PAMAP2 energy labels |
| 26 | IMUOrientation Estimation and Sensor Calibration for Wearables | Madgwick / complementary filter | IMU calibration logs |
| Fall Detection & Safety | |||
| 27 | FallThreshold-Based Fall Detection from Accelerometer | rule-based · Python | SisFall · FallAllD |
| 28 | FallML Fall Detection with ADL Rejection | RF / SVM / CNN · Python | SisFall · MobiAct |
| 29 | FallNear-Fall and Recovery Pattern Analysis | sequence models · Python | Fall datasets with recovery |
| 30 | FallPrivacy-Aware Fall Detection on Edge without Cloud Raw Data | TFLite · on-device | SisFall edge deploy |
| 31 | FallFalse Alarm Reduction Strategies for Real-World Wearables | post-filters · user feedback sim | Noisy free-living logs |
| 32 | FallMulti-Sensor Fall Detection: IMU + Barometer | feature fusion · Python | Fall sets with pressure |
| Sleep Monitoring | |||
| 33 | SleepSleep Stage Classification from Wearable EEG / PSG Subset | scikit-learn · CNN | Sleep-EDF · PhysioNet |
| 34 | SleepActigraphy-Based Sleep/Wake Detection | Cole-Kripke / ML · Python | MESA actigraphy concepts |
| 35 | SleepPPG-Based Sleep Staging Proxy from Wrist Wearable | deep models · Python | PPG sleep datasets |
| 36 | SleepApnea Event Detection Concepts from Wearable Signals | event detection · ML | Apnea-ECG · PhysioNet |
| 37 | SleepSleep Quality Score from Multi-Night Wearable Summaries | feature aggregation · report | Multi-night actigraphy |
| 38 | SleepCircadian Rhythm Metrics from Longitudinal Wearable Data | cosinor · Python | Longitudinal HR/actigraphy |
| HRV, Stress & Affect | |||
| 39 | StressHRV Feature Set for Stress vs Baseline Classification | NeuroKit2 · scikit-learn | WESAD |
| 40 | StressMultimodal Stress Detection: ECG + EDA + Accel | fusion ML · Python | WESAD |
| 41 | StressReal-Time Stress Index Streaming Demo | sliding HRV · thresholds | WESAD streaming sim |
| 42 | StressPersonalisation: Subject-Specific Stress Models | transfer / fine-tune · Python | WESAD leave-one-subject |
| 43 | StressEDA Peak Detection and Skin Conductance Response Analysis | NeuroKit2 · Python | WESAD EDA |
| 44 | StressAffect Recognition from Wearable Multimodal Streams | multimodal nets · PyTorch | WESAD · DEAP concepts |
| Edge ML, Hardware & Pipelines | |||
| 45 | EdgeTensorFlow Lite Model for On-Watch Activity Classification | TFLite · quantisation | UCI HAR · PAMAP2 |
| 46 | EdgeEdge Impulse Pipeline for Wearable Gesture / Activity | Edge Impulse · Arduino | Custom + public IMU |
| 47 | EdgeESP32 / Arduino Wearable Prototype: Sense → Feature → Classify | Arduino · BLE · sensors | Lab collected traces |
| 48 | EdgePower Profiling of Continuous Sensing vs Duty-Cycled Modes | current measurement · analysis | Prototype power logs |
| 49 | EdgeBLE Data Pipeline from Wearable to Phone Dashboard | BLE · Flutter/Python UI | Streaming session logs |
| 50 | EdgeModel Compression: Pruning / Quantisation for Microcontrollers | TFLite Micro · CMSIS-NN concepts | HAR / ECG tiny models |
| 51 | EdgeOn-Device Learning / Fine-Tuning Concepts for Personalisation | tiny training loops · Python | Subject adaptation sets |
| 52 | EdgeSensor Fusion Firmware Architecture for Multi-Modal Wearable | RTOS concepts · C/Python | Multi-sensor bench data |
| Clinical & Application Domains | |||
| 53 | AppRemote Cardiac Rehab Monitoring with Wearable HR Zones | HR zones · Python dashboard | Exercise ECG/PPG sets |
| 54 | AppElderly Activity and Fall Risk Score from Daily Wearable Use | risk scoring · longitudinal | HAR + fall feature sets |
| 55 | AppPost-Operative Ambulation Tracking with Wearable IMU | step/gait metrics · report | Clinical ambulation logs |
| 56 | AppDiabetes Lifestyle Logging: Activity + HR Context (No CGM Claim) | context features · diary fusion | Activity + self-report |
| 57 | AppSports Performance: Load and Recovery Proxies from Wearables | TRIMP / HRV recovery · Python | Athlete wearable logs |
| 58 | AppOccupational Heat / Fatigue Proxies from Wearable Sensors | temp + HR features | Occupational field sets |
| 59 | AppMedication Adherence Nudge System Triggered by Wearable Context | rule engine · notifications | Context event logs |
| 60 | AppPaediatric Activity Monitoring Ethics and Metric Design | literature + pilot metrics | Paediatric actigraphy concepts |
| Signal Quality, Validation & Ethics | |||
| 61 | ValCross-Validation Strategies for Subject-Independent Wearable ML | LOS O / group K-fold | Any multi-subject set |
| 62 | ValCalibration Drift and Long-Term Sensor Stability Analysis | drift metrics · Python | Longitudinal device logs |
| 63 | ValBenchmark Classical Features vs Deep Models on Same Wearable Set | ablation · sklearn + DL | PAMAP2 · WESAD · MIT-BIH |
| 64 | ValPrivacy: On-Device Inference vs Cloud Raw Stream Trade-offs | architecture comparison | Threat model notes |
| 65 | ValDemographic Fairness in Wearable ML Performance | subgroup metrics | Annotated subject metadata |
| 66 | ValRegulatory Awareness: Wellness vs Medical Device Claims | policy summary · citations | Guidance literature |
| Advanced & Capstone | |||
| 67 | AdvMultimodal Foundation Features for Wearable Time Series | self-supervised · PyTorch | Large wearable corpora |
| 68 | AdvFederated Learning across Wearable Users without Central Raw Data | FL frameworks · Python | Partitioned HAR / ECG |
| 69 | AdvDigital Twin of a Wearable User: Simulation of Sensor Streams | generative models · Python | Synthetic + real mix |
| 70 | AdvContinuous Authentication from Wearable Biometrics | ECG/gait biometric · ML | Biometric wearable sets |
| 71 | AdvAnomaly Detection for Rare Cardiac Events in Long Recordings | autoencoders · isolation | Long ECG Holter-style |
| 72 | AdvActive Learning to Reduce Labelling Cost for HAR / ECG | uncertainty sampling | Pool-based labelling sim |
| 73 | AdvOpen-Source Wearable Stack: Firmware + Cloud + Dashboard | ESP32 · MQTT · web UI | End-to-end lab demo |
| 74 | AdvComparative Study of Commercial Wearable APIs for Research | API sampling · metrics | Exported Fitbit/Garmin-style |
| 75 | AdvStress and Sleep Joint Modelling from Multi-Night Wearables | multi-task · longitudinal | WESAD + sleep proxies |
| 76 | AdvRobustness to Missing Sensors / Channels in Multimodal Wearables | dropout training · fusion | WESAD channel ablations |
| 77 | AdvReal-Time Dashboard: Live ECG/PPG/IMU Visualisation and Alerts | Streamlit · BLE · Python | Streaming session capture |
| 78 | AdvCurriculum: From Filtering to Deployed TFLite Model Documentation | full lab manual · code | Teaching package data |
| 79 | AdvMeta-Analysis Style Comparison of Published Wearable HAR Accuracies | literature table · replication | Public HAR benchmarks |
| 80 | AdvCapstone: End-to-End Wearable System — Sense, Process, Classify, Alert, Report | Python · TFLite · dashboard | Integrated multi-signal demo |
| 81 | AdvPPG + ECG Dual-Signal Fusion for Improved HR and Rhythm Context | multimodal fusion · Python | Simultaneous ECG-PPG sets |
| 82 | AdvExplainable Fall Detection: Feature Importance for Clinician Trust | SHAP · rule extraction | SisFall with explanations |
Datasets are public research sources (PhysioNet, MIT-BIH, WESAD, PAMAP2, SisFall, Sleep-EDF, PPG-DaLiA, etc.). Always respect licences and ethical use. Contact us for pipelines, metrics, university-format report, PPT and viva Q&A.
Why Choose Us for Wearable Biomedical Projects?
Bangalore-based guidance for BE, BTech and MTech biomedical and ECE students.
ECG & PPG
QRS detection, arrhythmia classification, HR/SpO2 estimation and motion-robust pipelines.
IMU & Fall
Activity recognition, gait features and fall detection with edge-ready models.
Sleep & Stress
HRV, multimodal stress (WESAD) and sleep staging from wearable-compatible signals.
Edge & Capstone
TFLite, Edge Impulse, ESP32 prototypes and full sense-to-dashboard systems.
FAQ — Wearable Biomedical Projects
Wearable Biomedical Lab — Bangalore
Signal processing, ML and edge deployment support for wearable health projects.
Arrhythmia
HR / SpO2
HAR / Gait
Detection
HRV Stress
Edge Deploy
Prototypes
Support