Dalton Simancek

NLP Research Specialist (Graduate Researcher)

• Fine-tuned and validated a BERT-based de-identification NLP system with the lead researcher at Michigan Medicine. This system facilitated the automated removal of 18 HIPAA identifiers from unstructured clinical text fields extracted from thousands of patients' electronic health records
• Developed name pseudonymization NLP post-processing procedure to reduce leakage of sensitive patient names by 70% through obscuring the visibility of de-identification system false negatives at scale
• Implemented a novel NLP fairness evaluation framework utilizing patient demographic information to assess de-identification performance gaps across patient subpopulations. Proposed a stratified patient record sampling workflow to address equity gaps in system performance

Data Scientist

• Operationalized a new data platform within the HomeKit framework, among the first to develop ETL and analysis workflows from novel cloud infrastructure. Processed device logging from millions of Apple and 3rd-party accessories, delivering database aggregates, dashboards, and Python modeling scripts to support QA and Product KPIs
• Forecast impact of Product and QA development efforts: facilitating hardware depreciation, mapping code changes to software termination rates, and projecting trends in customer acquisition and maturity
• Collaborated with privacy engineers to write and approve 50 new data fields into 10+ HomeKit software updates, balancing the utility of information collection with customer privacy risks
• Delivered reports on HomeKit growth to the finance team and Apple executives for eight quarterly earnings calls

Data Analyst

• Led engineering quality assurance strategies to mitigate software terminations (e.g., app crashes, kernel panics) on the latest watchOS builds. Analyzed device logging and established termination rate goals across millions of Apple Watches, resulting in record-low termination rates across three major concurrent and over 15 minor watchOS updates.
• Championed watchOS stability goals through weekly communication with dozens of watchOS engineers and executives regarding top crashes and termination rates of 3+ software updates at a time, strategically managing software stability through the development cycle from internal testing to Beta and Public releases.

Research Associate

• Facilitated the collection and analysis of data for grants and publications among three principal investigators via surveys, interviews, and focus groups.
• Created 30+ infographics and educational videos to educate physicians, patients, and parents about diagnostic testing, patient experiences, and treatment curriculums related to newborn screening, adolescent diabetes, and pediatric weight management.

   Grants

In My Shoes: Using Parent Narratives as a Newborn Screening Educational Tool
2014 Challenge Award Recipient, Baby's First Steps, Genetic Alliance
Web-based catalogue of parents’ self- narratives about their experiences receiving a positive newborn screening result

   Publications

Goldenberg AJ, Ponsaran R, Gaviglio A, Simancek D, Tarini BA. Genomics and Newborn Screening: Perspectives of Public Health Programs. International Journal of Neonatal Screening. 2022; 8(1):11. doi: 10.3390/ijns8010011

Scherer LD, Finan C, Simancek D, Finkelstein JI, Tarini BA. Effect of "Pink Eye" Label on Parents' Intent to Use Antibiotics and Perceived Contagiousness. Clin Pediatr (Phila). 2016 06; 55(6):543-8. doi: 10.1177/0009922815601983.

Global Information Engagement Intern

• Developed a web-based solution to support client intake data collection and storage for a Cape Town non-profit supporting the blind and visually impaired

User Experience Intern

• Evaluated the usability, usefulness, and acceptability of a software platform to report workplace incidents; the assessment featured interaction maps, user interviews, usability tests, heuristic evalutions and surveys

Information Intern

• Planned a set of KPIs to monitor and report incubator health

Citizen Interaction Design Intern

• Designed prototype for BREADCRUMBS NATURE GUIDE*: a bluetooth-beacon powered Android App to help the blind and visually impaired navigate nature trails

   *Featured at CES 2016 STUDENT SHOWCASE

NYC Innovation Trek

• Created and workshopped a pitch for 'Troup': a mobile platorm to plan, coordinate and share group travel to a set of companies across Manhattan

Community Video Production Intern

• Filmed and Produced a series of short documentaries for marketing the services of a daytime homeless shelter

Disney College Program Intern

• Performed custodial services and street entertainment at the Magic Kingdom Theme Park

University of Michigan Medical School

Coursework and Dissertation Interests

NLP for Health Data | Deep Learning Architectures | Computational Bioinformatics | Distributed Learning | Differential Privacy | Implementation Science Frameworks

University of Michigan School of Information