Use of electrical impedance spectroscopy (EIS) for diagnosis and prediction of reproductive status in female swine

Efficient management of fertility within the swine breeding herd is vital for maximizing productivity. Key reproductive events, such as estrus detection, pregnancy confirmation, and farrowing, must be effectively managed to optimize animal flow and pig production. Currently, few tools are available to accurately and quickly diagnose and predict reproductive events and instead rely upon labor to identify the outcome. Electrical impedance spectroscopy (EIS) employs a range of frequencies to measure changes in tissues responses to an applied charge. The objective of this research was to evaluate EIS as a tool for diagnosing and predicting reproductive events in swine. Impedance readings were obtained intravaginally using an EIS device equipped with a four-electrode transducer and an internal processor, measuring tissue impedance (Ω) and phase across a frequency range of 1000-29000 Hz and 10-400 ohms. The device was connected to a mobile unit for wireless data collection and storage. Prior to scanning each sow, the device was cleaned and disinfected before insertion into the vagina. Once in the vagina, readings were obtained within ~15 seconds. One sow could be scanned every 90 seconds when in adjacent stalls. Experiment 1 tested EIS to identify estrus in weaned sows (n=546) that were scanned from the day before weaning until day 6 post-weaning. Experiment 2 tested EIS as a method for early determination of pregnancy with scanning on days 1, 10, and 19 in weaned sows (n=500) following insemination. Experiment 3 used EIS to predict the day of farrowing in sows (n=503) from day 113 of gestation until the day of farrowing. Following the collection of data, EIS readings were automatically uploaded to a dedicated website for processing and analysis. Data were analyzed in RStudio and Python to develop and evaluate impedance patterns by day and models for the prediction of reproductive events. Our results indicate that the prediction and identification of estrus within 36 hours in weaned sows had an F1 score = 0.89 for no estrus and 0.76 for estrus detection, with a balanced accuracy of 0.85. For diagnosing pregnancy, the model attained an F1 score = 0.86 for non-pregnant sows and 0.52 for pregnant sows, with a balanced accuracy of 0.78 on day 19. For farrowing, the model reached an F1 score = 0.88 for farrowing not occurring and 0.70 for farrowing occurring within the next 36 hours, with a balanced accuracy of 0.83. Our results suggest that various frequencies can detect changes in the vaginal impedance in the days before a reproductive event. Our efforts continue in impedance pattern analysis by day and frequency as well as developing predictive and machine learning models to improve the accuracy and potential for practical use of this technology in swine herd management.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Animal Sciences

Grantor

University of Illinois Urbana-Champaign

Year dc:date

2025

Author dc:creator

• Vilchez Herrera, Pastor

Contributors dc:contributor

• Knox, Robert V
• Wheeler, Matthew
• Miller, David
• Bhalerao, Kaustubh

• impedance
• prediction
• estrus
• pregnancy
• farrowing
• management

Statement dc:rights

• ©2025 by Pastor Vilchez. All rights reserved.

Language dc:language

en, eng

Handle dc:identifier

https://hdl.handle.net/2142/129168

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F129168