Areas of Interest
Below are several broad research domains that align with CACoM's goals. Each connects to real-world biomedical and clinical challenges where computational, statistical, or experimental work can make a measurable contribution.
These are not predefined topics — they are invitations to explore. You are expected to refine any idea into a specific, feasible, and testable research question, and to demonstrate access to the required data, devices, or collaborators.
These areas are conceptual entry points, not ready-made projects. Your proposal must clearly define your own refined topic, data, method, and evaluation strategy, and show how you will make the work feasible within one semester.
👶 Fetal Heart Rate, CTG, and Doppler Research​
Fetal heart rate (FHR) monitoring remains one of the cornerstones of obstetrics — yet its accuracy, interpretation, and clinical value are still debated. Modern CTG (cardiotocography) systems primarily rely on Doppler ultrasound, which estimates fetal heart motion by detecting tiny frequency shifts in reflected sound waves. While this method is widely used, signal quality, device validation, and algorithm transparency remain open research challenges.
We have access to an in-house fetal heartbeat simulator, which can reproduce realistic mechanical pulse signals. This allows for both experimental and computational projects — from testing devices to developing signal algorithms and data analytics.
Possible directions (illustrative only):
- Studying how ambient noise, acoustic interference, or probe placement affect Doppler signal quality.
- Designing or benchmarking algorithms for beat detection or heart rate variability (HRV) computation.
- Using public CTG datasets or simulated Doppler signals for reproducibility and machine-learning studies.
- Exploring transformations between ECG- and Doppler-derived series (e.g., smoothing or downsampling to mimic CTG resolution) — provided you can demonstrate data access.
- Investigating novel quantitative or ML-based metrics of variability, noise robustness, or clinical interpretability.
- Developing or testing research-grade Doppler devices to enable open validation pipelines (not medical devices).
Some of these directions require hardware access or specialized datasets.
Your proposal must specify how you plan to obtain, simulate, or share data responsibly.
Relevant skills:
Signal processing, machine learning, time-series analysis, experimental design, numerical modeling, and statistical evaluation.
🦶 Movement, Balance & Wearable Sensors​
The biomechanics of human movement — especially gait and posture — provide rich opportunities for quantitative analysis. Using wearable sensors such as IMUs (inertial measurement units; e.g. actibelt®) and foot-sole pressure sensors, we can capture motion data in both laboratory and real-world environments.
Our setup allows experiments with BLE-enabled barefoot insoles and body-mounted IMUs, and access to selected actibelt® datasets, including those collected in extreme conditions such as microgravity or spaceflight. Some datasets are unlabeled and may require creative preprocessing or self-supervised approaches.
Possible directions (illustrative only):
- Comparing barefoot vs. minimalist footwear walking and its effects on sensory feedback or balance.
- Developing or validating algorithms for speed, step length, or odometry estimation from foot or waist-mounted IMUs.
- Investigating speed estimation and related algorithms (treadmill, free walking etc.).
- Recognizing posture transitions or activity segments from IMU data streams.
- Exploring drift correction, sensor fusion, or context-aware constraints for improved position estimates.
If your project involves data collection, specify your experimental setup (sensor type, sampling rate, participant safety). If using actibelt® or other clinical data, discuss data access and labeling limitations early with the instructors.
Relevant skills:
Signal processing, sensor fusion, time-series modeling, experimental design, and algorithm validation.
😷 Flexeal: Smart Patch for Mask Sealing & Sensing​
Flexeal is a silicone patch designed to seal FFP2 masks at the nose bridge to prevent air leakage toward the eyes. Its soft, adhesive surface also makes it an ideal base for embedding micro-sensors, such as pressure, temperature, or gas sensors — opening possibilities for wearable monitoring and comfort optimization.
Possible directions (illustrative only):
- Characterizing seal performance under different face geometries, movements, or humidity conditions.
- Embedding miniature environmental sensors (e.g., COâ‚‚, humidity, thermal gradient) to study breathing patterns or mask comfort.
- Modeling airflow and leak dynamics using experimental or computational data.
- Designing data pipelines for sensor calibration and quality control.
Relevant skills:
Sensor integration, experimental measurement, signal processing, mechanical modeling, and low-power electronics.
🎤 Acoustic & Physiological Signals​
Acoustic sensing provides a non-invasive window into human physiology. From electronic stethoscopes to respiratory microphones, acoustic signals can capture complex biological processes — but interpreting them reliably is a challenge.
Possible directions (illustrative only):
- Extracting and comparing spectral or temporal features from heart or lung sounds.
- Evaluating algorithms for detecting pathological signatures (e.g., murmurs, respiratory irregularities) in controlled or open datasets.
- Investigating signal quality metrics, denoising, or automated screening methods.
- Studying how physiological variability (e.g., breathing depth, posture) affects recordings.
Acoustic projects require clean, well-labeled recordings. If collecting your own data, describe the recording environment, equipment, and calibration in your proposal.
Relevant skills:
Audio signal analysis, feature extraction, machine learning, and validation methodology.
🩺 Surveys, Systematic Reviews & Clinical Data Interpretation​
Not all CACoM projects are technical — some explore how clinicians use, interpret, or trust data-driven tools.
These can include structured literature reviews, survey-based investigations, or meta-analyses — provided they are rigorous and quantitative.
Possible directions (illustrative only):
- Conducting a systematic review of ML methods used in obstetric monitoring or wearable sensing.
- Designing and analyzing a survey on algorithm interpretability or data transparency among clinical researchers.
- Quantitatively reviewing reproducibility statements and dataset availability across computational medicine papers.
- Analyzing how fetal monitoring or wearable-sensing studies report uncertainty, validation metrics, and dataset limitations in their predictive models.
For survey studies, include a plan for recruitment, dissemination, and ethical handling of responses. Approval may be required for any human-participant data.
Relevant skills:
Survey design, statistical analysis, data visualization, systematic review methodology.
🧩 Other Directions & Custom Proposals​
Not every strong idea fits neatly into existing categories. However, projects outside the listed areas must demonstrate exceptional initiative, domain understanding, and motivation. Unless your project originates from an approved external collaborator, it must convincingly justify its clinical relevance, computational challenge, and evaluation strategy.
Examples include:
- Projects proposed by clinicians, research groups, or industry partners.
- Analytical studies that integrate multiple data sources (e.g., sensors + public datasets + modeling).
- Theoretical or simulation-based work that addresses biomedical mechanisms or model validation.
Custom proposals are welcome but held to a higher bar of justification. You must show deep understanding, a realistic plan, and strong alignment with CACoM's scientific and educational goals.
General Notes​
- These domains are starting points, not fixed projects.
- Your proposal must define a specific question, dataset, and evaluation plan within one of them.
- You are encouraged to combine areas creatively — e.g., IMU sensors + physiological modeling, or acoustic data + machine learning.
- Hardware or software development is acceptable only when it supports a clear clinical or scientific goal.
One of the exciting things about CACoM is that the world is your oyster — you can use (almost) any modern tool, framework, or technology you like. But remember: these tools are means to an end, not the end itself. Do not propose projects that are purely about tool building (apps, dashboards, GUIs) without scientific evaluation. CACoM emphasizes understanding and quantifying clinical or physiological phenomena, not product development.