Contributed Poster Presentations: Section on Medical Devices and Diagnostics

Artificial intelligence (AI) doctors are chatbots that diagnose patients on the Internet using algorithms to determine patients' condition and need for medical treatment. I build a mathematical model to investigate how AI doctors affect patients and the efficiency and utilization of healthcare. I hypothesize that AI doctors improve patients' healthcare utilization decisions, alleviate healthcare access inequity, improve healthcare efficiency and patients' well-being. I also hypothesize that more trust in AI doctors benefits patients. My analysis suggests that the hypotheses hold only when AI doctors' diagnosis is very accurate. When AI doctors are not very accurate, their impacts on patients and healthcare vary with patient types (the severity of their medical condition and the cost they face to access healthcare). Specifically, AI doctors change patients' decisions to go to hospitals when patients are moderately uncertain about their need to go there. AI doctors alleviate healthcare inequity by increasing healthcare utilization by patients with low access and discouraging utilization by patients with easy access. AI doctors benefit patients with moderate access to hospitals, sever 

Pancreatic cancer (PC) remains one of the most lethal cancers due to challenges with early diagnosis. Timely identification of PC is essential for improving prognosis and survival rates. In 2024, the National Cancer Institute reported 51,750 deaths from PC, making up 8.5% of all cancer deaths. Through a meta-analysis we aim to systematically compare the specificity and selectivity of various diagnostic tools, including biomarkers, imaging techniques, and biopsies.
Specificity is a methodology's ability to identify a true negative (correctly not diagnosing an individual with PC), while selectivity is the ability to identify a true positive (correctly diagnosing an individual with PC). Cancer Antigen 19-9 is common as a pancreatic cancer biomarker to improve screening effectiveness. Positron emission tomography imaging detects metabolic activity in tumors through radiotracers. Fine-needle aspiration biopsies are a minimally invasive approach to obtain tissue. Also, to improve diagnostic accuracy these methodologies, along with others, are used in conjunction. The goal is to determine which is most effective for diagnosing PC, and which can be used to diagnose PC at earlier stages. 

As researchers transition from SAS to R, understanding the differences between SAS PROC MIXED and R's lme4::lmer is crucial for accurate linear mixed-effects modeling. Both tools implement linear mixed-effects models, yet users may encounter mismatches in outputs due to the distinct optimization algorithms employed by PROC MIXED and lmer. This poster will also explain how to handle different warning and error messages in lmer. By highlighting these key differences and providing practical guidance, this poster aims to support users in effectively adopting lme4 in R for their linear mixed-effects modeling needs.