Lucy Spicher

SUMMARY: Data scientist and mechanical engineer specializing in biomedical signal processing, machine learning, and wearable physiological sensing for digital health. Completed a PhD developing time-series algorithms and machine learning classification models for fetal movement detection using wearable inertial sensors and currently consulting for an early-stage digital health startup developing next generation acoustic and motion-based fetal monitoring systems. Experienced in data-driven algorithm development, multimodal physiological data analysis, statistical modeling, and translating findings into product-oriented solutions through cross-functional collaboration.

SKILLS: Programming (Python/Jupyter, MATLAB, R/RStudio) | Machine Learning (PyTorch, TensorFlow, Scikit-Learn, Classification, Deep Learning, CNNs, LSTMs, Random Forests, Model Evaluation) | Statistical Modeling (Linear mixed-effects models, Regression) | Signal Processing (Data Visualization, Filtering, Sensor Fusion, Feature Engineering, Time-series Analysis) | Wearable Technologies (Inertial Measurement Units, Acoustic Microphones) | Research Methods (Human-Subjects Research, Study Design, Protocol Development, IRB Processes, Scientific Communication, Manuscript Development)

PROFESSIONAL EXPERIENCE

Research Scientist, Wearable Sensing (Consultant) – Project Stardust

April 2026 – Present

• Lead development and evaluation of inertial sensing algorithms for a wearable fetal monitoring platform integrating acoustic and motion sensing

• Design reproducible signal processing pipelines for synchronized multimodal physiological data, automated data quality checks, and pilot data review

• Collaborate across hardware, firmware, and data science efforts to evaluate sensor count, placement, specifications, and data quality for human-subject studies

• Translate physiological sensing requirements into practical engineering and study design decisions for proof-of-concept development

Graduate Research Assistant – University of Michigan

August 2021 – May 2026

•  Led IRB-approved human-subject studies investigating wearable physiological sensing for fetal movement detection, integrating study design, data collection, signal processing, statistical modeling, and machine learning

• Developed and evaluated multimodal machine learning classification models using accelerometer and gyroscope data that improved fetal movement detection sensitivity by 6% over single-sensor approaches; Random Forest models achieved the highest performance (AUROC = 0.81) and outperformed CNNs under data-limited conditions

• Applied linear mixed-effects models to characterize fetal movement and maternal motion artifacts, using the resulting estimates to establish data-driven filtering and artifact-rejection parameters for wearable sensing

• Evaluated machine learning algorithms using multiple ground-truth and validation strategies, demonstrating that ultrasound-derived labels improved specificity by 7% over maternal perception and that subject-independent validation reduced AUROC from 0.80 to 0.61

• Benchmarked BiLSTM models against hand-engineered feature-based models for IMU-based gait screening, demonstrating comparable performance between minimally processed time-series (AUROC = 0.86) and traditional feature-based approaches (AUROC = 0.89)

• Built reproducible Python pipelines for physiological time-series data processing, quality assessment, feature engineering, statistical analysis, and machine learning model evaluation, translating findings into peer-reviewed publications and international conference presentations

• Developed a validated contextual engineering design framework from interviews with 58 engineering practitioners, producing 9 context categories and 29 contextual factors to guide future engineering design

FIRST AUTHOR PUBLICATIONS

1. Spicher L, Bell C, Sienko KH, Huan X. Comparative Analysis of Machine Learning Approaches for Fetal Movement Detection with Linear Acceleration and Angular Rate Signals. Published in MDPI Sensors 2025. <https://doi.org/10.3390/s25092944>

2. Spicher L, Kong L, Avaala J, Batool A, Burleson G, Toyama K, Sienko KH. Contextual Factors Incorporated by Experienced Design Practitioners: A Study Across Three Design Domains. Under review.

3. Spicher L, Bell C, Sienko KH, Huan X. Four Data-driven Design Choices for IMU-based Fetal Movement Detection: Sensitivity, Thresholding, Windowing, and Filtering. Proceedings of the 2026 Design of Medical Devices Conference. < https://doi.org/10.1115/DMD2026-1071>

4. Spicher L, Bell C, Huan X, Sienko KH. Exploring Random Forest Machine Learning for Fetal Movement Detection using Abdominal Acceleration and Angular Rate Data. Proceedings of the 2024 IEEE Engineering in Medicine and Biology Conference (EMBC). <https://doi.org/10.1109/EMBC53108.2024.10782456> 

5. Spicher L, Bell C, Wiens J, Sienko KH, Huan X. Methodological Evaluation of Label-Source Selection and Evaluation Strategies on Reported Performance in Wearable Fetal Movement Detection. Manuscript in preparation.

Additional co-authored publications available upon request

EDUCATION

University of Michigan (Ann Arbor, MI)

Ph.D. & M.S. in Mechanical Engineering - August 2026

GPA: 4.00/4.00

Pennsylvania State University (University Park, PA)

B.S. in Mechanical Engineering & B.S. in Biomedical Engineering - May 2021

GPA: 3.71/4.00

GLOBAL HEALTH INNOVATION EXPERIENCE

Graduate Trainee – NIH Acute and Critical Care in Engineering Training Program (ACCETP)

August 2023 – 2024

• Completed immersive clinical shadowing and training in clinical workflows, needs identification, and healthcare technology commercialization for translational medical technologies

Clinical Needs Assessment Intern – Global Health Design Initiative

June – August 2023

• Conducted clinical needs assessments across rehabilitation clinics in India, identifying 100+ unmet needs and translating clinical observations into engineering design requirements

Product Development Lead, Occupational Therapy Tools – Kijenzi

August 2017 – May 2021

• Led needs-finding across seven rural Kenyan clinics, resulting in engineering specifications for 17+ 3D-printed rehabilitation devices

OTHER PROFESSIONAL EXPERIENCE

CrossFit Level 2 Trainer – Fit2Live Gym

October 2021 – June 2026

• Led coaching and individualized training for a community of 150+ members while designing personalized performance programs based on movement analysis and measurable progress

• Programmed and hosted structured 15-hour youth fitness camps focused on goal tracking and skill development