On 17th of November, 2026, at 15.00 in person, at the University of Latvia (Raiņa bulvāris 19) Auditorium 20, as well as remotely, the UL Promotional Council of Computer Science and Informatics, and Electrical Engineering, Electronics, Information, and Communication Technologies will hold an open meeting, during which Didzis Lapsa will defend his doctoral thesis (Ph.D) in Engineering and Technology.

Topic: "Analysis of heart rate estimation via enhanced wearable bioimpedance device".

Scientific supervisor: Dr. sc. comp. Atis Elsts.

Abstract:

Heart rate (HR) estimation is an essential aspect of healthcare and fitness, with wearable devices being widely adopted for this purpose. Despite their growing popularity, existing HR estimation techniques often face challenges such as high power consumption, limited accuracy, and reduced long-term user comfort. This thesis explores the use of bioimpedance (BioZ) measurement as an alternative method for HR estimation, particularly suited for wearable devices. The work presents three main contributions: 1. a wrist-worn low-power BioZ signal measurement platform (WearEdi V2), achieving sub-milliwatt sensor operation under duty-cycled conditions; 2. a wavelet ridge–based signal-to-noise ratio (SNR) metric specifically designed for non-stationary and non-Gaussian

BioZ signals (protected by Latvian national patent No. 15878 B1); and 3. The “BPM DL” two-stage HR estimation algorithm, combining frequency-domain cross-correlation with adaptive time-domain peak refinement. Experimental validation across 14 test subjects and multiple electrode configurations demonstrates that the proposed algorithm reduces the mean absolute percentage error (MAPE) of HR estimates by 50.4 % compared to a prominence-based peak-detection baseline. With optimized bipolar electrode configurations, MAPE values as low as approximately 8.0 % are achieved, with BPM differences in the range 0.059–0.467 BPM under high-quality signal conditions. The findings indicate that bioimpedance measurement can provide viable HR estimation, offering a complementary approach to existing techniques. The results have been disseminated through five peer-reviewed SCOPUS-indexed publications and one granted Latvian national patent.

Reviewers:

  1. ‎Dr. sc. ing. Jānis Grabis, University of Latvia;
  2. Dr. sc. ing. Andis Supe, Riga Technical University;
  3. Dr. Daniel Teichmann, University of Southern Denmark, Denmark.

The doctoral thesis is available at the LU Library, Raiņa bulvāris 19.

Participation in the online session is by prior application, writing to sintija.silina@lu.lv by 12th of November.