Exploring Discrepancies: Analyzing Electronic Medical Records Data Against Direct Observations; Ethical Considerations and Care-Centric Integration of AI-Diagnostics in Healthcare

Author: ORCID icon orcid.org/0009-0006-3299-4857
Park, Sammy, School of Engineering and Applied Science, University of Virginia
Advisors:
Earle, Joshua, EN-Engineering and Society, University of Virginia
Riggs, Robert, EN-SIE, University of Virginia
Abstract:

Technical Project
The objective of this project is to optimize patient and worker experiences at primary care clinics, with a focus on the University Physicians Clinic of Charlottesville (UPC). Through qualitative observations conducted in two-hour increments, the team collected data on appointment milestones such as patient entry, physician-patient interactions, and patient departure. This observational dataset aimed to align with the clinic's electronic medical records (EMR) data, however, significant disparities emerged between the two datasets, particularly regarding appointment durations and physician-patient interactions.
The UPC Clinic, facing challenges after the COVID-19 pandemic and current nursing shortages, relies on EMR data for patient scheduling and clinic-based decision-making. Past studies focused on patient flow analyses and EMR data assessment which laid the groundwork for this project. However, our in-person observations revealed large discrepancies in EMR data accuracy which may be impacting informed decision-making and patient flow optimization.
The study's methodology is structured into three main parts. Part 1 involved observational data collection where the team shadowed nurses to capture patient flow processes. Part 2 compared observational timestamps with EMR records which revealed inconsistencies in appointment durations and physician-patient interactions. Part 3 analyzed EMR data variance. Our findings drove our recommendations that centers on the need for improved data collection and cleaning procedures.
To begin our analysis, we cleaned the EMR dataset to mitigate the influence of outliers and missing data on our statistics. However, disparities persisted. Therefore, for future reference, additional investigation and refinement of data collection methods will be necessary to identify the underlying issues. Initial analysis also revealed that physician appointment capacities are being overestimated due to frequent appointment delays.
Our analysis of observational and EMR data highlighted significant discrepancies in 20 and 40 minute appointment length durations and the time a physician spends with a patient during an appointment. In our observations, the average appointment duration across all visit types exceeded the recorded data in the EMR. Additionally, inaccuracies persisted in the EMR data despite undergoing a data cleaning process. Variations in the way nurses and physicians deliver services also contributed to data inconsistencies, so the team recommends implementing standardized procedures without shortcuts.
Overall, our technical project at the UPC of Charlottesville showed substantial disparities between observational findings and EMR data. Despite our attempts to align with EMR records, persistent discrepancies highlighted challenges in data accuracy and decision-making. We identified the overestimation of physician appointment capacities, along with frequent appointment delays. Our recommendations to overcome this challenge focuses on improved and transparent EMR data collection methods as well as standardized nursing and physician procedures. Addressing these issues will prove helpful for improving patient care, optimizing clinic operations, and minimizing the impact of data inconsistencies on informed decision-making.

STS Project
Through this paper, I delve into the ethical aspects of integrating artificial intelligence (AI) diagnostics in healthcare and explore its potential to enhance rather than replace human care in healthcare. I aim to explore how AI-diagnostics can improve patient care while considering ethical factors such as data handling, algorithmic fairness, and patient well-being. To investigate AI-diagnostic integration in healthcare, I conducted a comprehensive literature review using academic databases like IEEE Xplore and JSTOR. I also incorporate care ethics as the guiding framework into my study, which puts emphasis on the importance of care in moral decision-making.
When analyzing the benefits of AI-diagnostics in healthcare, I highlight its impact on efficiency, accuracy, and innovation. I explore how AI diagnosis technologies streamline healthcare processes, enhance diagnostic accuracy, and drive innovation in personalized treatment plans, which all ultimately improve patient treatments and well-being. I discuss key ethical concerns and values, including data privacy and security, bias and fairness in algorithms, and transparency and accountability in decision-making processes. I also examine studies centering on issues such as patient data protection, the mitigation of algorithmic biases, and ensuring transparency to maintain trust in AI diagnosis technologies.
My findings point to the importance of responsible AI integration in healthcare and push for transparency, bias mitigation, and collaboration among stakeholders in the healthcare actor-network. I also encourage for ethical standards that align with care ethics to guide AI-diagnostic deployment, so that AI in healthcare systems prioritizes patient-centered care and upholds public trust. Overall, I provide insights into the functional advantages and ethical considerations of AI-diagnostic implementation in healthcare. By using a responsible and patient-centric approach, healthcare professionals and policymakers can leverage AI-diagnostics to benefit patients while safeguarding their rights and ensuring equitable access to healthcare services.

Degree:
BS (Bachelor of Science)
Keywords:
Electronic Medical Records (EMR), Patient Flow Observational Data, Data Variance Analysis, Artificial Intelligence (AI) Diagnostics, Healthcare Integration, Care Ethics, Patient-Centered Healthcare
Notes:

School of Engineering and Applied Science

Bachelor of Science in Systems Engineering

Technical Advisor: Robert Riggs

STS Advisor: Joshua Earle

Technical Team Members: Sammy Park, Grace Fitzgerald, Catherine Irons, Colin Miehe, Avery Schebell

Language:
English
Rights:
All rights reserved (no additional license for public reuse)
Issued Date:
2024/05/09