Development and Quantitative Evaluation of a mmWave Contactless Vital-Parameter Monitoring Prototype with Home Assistant Integration for Homecare Support

Dominik Mayrhofer, BSc.
dh231810@ustp.ac.at
Master Digital Healthcare, St. Pölten University of Applied Sciences 2026

Aim and Research Question(s)

This thesis presents a contactless vital-parameter monitoring prototype based on a 60GHz mmWave radar sensor integrated into Home Assistant. The aim was to evaluate technical feasibility, measurement behaviour, and initial usability in a controlled, non-clinical proof-of-concept setting.

  • How accurate and reliable are the measured vital parameters compared to standard reference devices in a lying position?

  • How does the system perform under movement, blanket coverage, or changes in breathing habits?

  • What requirements and design principles should guide a userfriendly dashboard for homecare professionals to support effective usage and decisionmaking?

Background

Continuous monitoring of vital parameters such as heart rate and respiratory rate is discussed to support early detection of complications and to relieve both professional and informal caregivers, particularly in home-based settings [1], [2]. Recent work highlights 60GHz mmWave radar as a promising contactless approach for detecting respiration- and heartbeat-related chest wall movements without audio or visual data, supporting privacy-preserving use across typical domestic conditions [2].

Methods

The prototype used a Seeed Studio MR60BHA2 sensor kit with an ESP32C6 microcontroller and provided per-second HR and RR outputs via ESPHome to a local Home Assistant instance. A quantitative pilot study (N = 5) was conducted at “Diakoniewerk Gallneukirchen” with four scenarios (relaxed, blanket coverage, movement, breath hold). Vital parameters were compared against a Braun Pulse Oximeter 1 and a Garmin Venu 3. Data were time-aligned to one-second resolution, cleaned of predefined outliers, and analysed using Bland–Altman analysis on complete per-second pairs. Usability was assessed with PSSUQ v3 and open-ended questions.

Results and Discussion

Heart rate (HR): Pulse oximeter against mmWave showed a mean bias of −7.19 bpm with 95% LoA −38.35 to 23.97; Garmin against mmWave showed a mean bias of −7.23 bpm with 95% LoA −38.11 to 23.65.
Respiratory rate (RR): Garmin against mmWave showed a mean bias near zero (0.16 brpm) with 95% LoA −7.79 to 8.09. Time-series plots indicated inconsistent data availability and more irregular trajectories during movement, and a consistent RR drop towards values near zero during breath hold was not observed across participants.
Usability: Overall PSSUQ mean score was 1.18 (SD 0.19) and the participants described the dashboard as clear.

Conclusion

The prototype demonstrates functional integration of contactless sensing into a local smart-home platform, but measurement variability and inconsistency limit practical interpretability. Future work should prioritise output stability, simplified installation for non-technical users, and larger evaluations with longer observations and multi-person presence. AI or machine-learning-based integrations could also be explored as an additional option.

References

[1] Yuvraj. “How touchless patient monitoring is defining future of healthcare landscape” [2] S. Jiang and B. Hettich. “Contactless measurement of vital signs with radar sensors”