5 articles found

Building the TacMed Chatbot to Support Medical Education in Low-Resource Settings: A Low-Code Platform Approach

Abstract

Building the TacMed Chatbot to Support Medical Education in Low-Resource Settings: A Low-Code Platform Approach

16 June 2025

Introduction As technology continues to advance, humans and technology have developed together to meet our needs, changing the way we live. This evolution forms the foundation of cyberpsychology [1]. Chatbots, which are virtual agents that communicate with users, play an important role in fields like customer service, healthcare, and e-commerce [2,3]. Chatbots can either follow simple rules (rule-based) or use AI technologies like Natural Language Processing (NLP) and Machine Learning to interact more intelligently. However, AI chatbots may present challenges, including the risk of delivering incorrect information, especially in high-stakes domains like healthcare. Health Tech Without Borders (HTWB) developed a TacMed Chatbot for frontline responders in Ukraine. The chatbot provides lifesaving medical information in critical situations. This study describes the steps taken to build this chatbot using a low-code platform, addressing the rationale behind the design choices and the process of deployment. Methods Pre-planning and Research The TacMed Chatbot project began by assessing the needs of Ukrainian frontline responders. The results highlighted the need for a simple, protocol-based tool, as many first responders in conflict zones had limited medical education and worked in high-stress conditions. A decision was made to use a low-code platform to develop the chatbot, focusing on decision tree logic. A low-code platform is a software development approach that requires minimal coding, enabling rapid application development using visual interfaces. This makes it accessible and allows for faster deployment of applications. The TacMed Chatbot was built using a low-code platform to streamline development and localization. Decision tree logic is a structured rule-based model that guides users through predefined steps based on their inputs, ensuring consistency and reliability Chatbot Design The chatbot used vetted information from sources such as Stop the Bleed and Tactical Combat Casualty Care (TCCC) protocols. A low-code platform was selected to facilitate decision tree logic and ensure seamless localization into multiple languages. Key design considerations included adapting the chatbot for Ukrainian medical terminology and integrating visual aids to support understanding. Testing Phase The testing phase involved diverse user groups, including medical professionals and civilians, across the US, EU, Ukraine, and Middle East. The team ran feedback sessions and tested the chatbot’s clarity and usability during a conference in Lutsk, Ukraine and virtually to medical professionals in Sudan. The goal was to ensure the chatbot was simple, effective, and appropriate for both healthcare professionals and non-medical users. Results By the end of July 2023, the TacMed Chatbot had delivered over 32,000 messages to more than 500 users. Its primary use cases involved providing emergency medical protocols for treating war casualties. Feedback from users indicated the chatbot's ease of use and its potential to be a valuable educational tool for frontline workers. Discussion The medical field continues to embrace chatbots, leveraging digital technologies to drive economic growth and improve public services. The TacMed Chatbot is an example of how technology can be harnessed to support healthcare delivery in conflict zones. The development process highlighted the benefits of using a low-code platform [4], which allowed for efficient creation, localization, and easy adjustments. Despite not utilizing AI, the TacMed Chatbot proved effective in delivering critical medical information. It complemented other educational tools and demonstrated the potential for chatbots to enhance long-term memory retention for medical protocols. However, further developments could integrate AI to handle unrecognized queries, enhancing the chatbot's ability to provide broader support. Limitations The TacMed Chatbot, while effective, does not incorporate advanced AI, limiting its ability to handle complex or out-of-scope queries. Moreover, it was designed primarily for frontline workers in Ukraine and Sudan, and its adoption may face challenges in other regions due to language and cultural differences. Additionally, its reliance on decision tree logic may restrict flexibility in responding to unique situations outside the predefined protocols. Conclusion and Future Directions The TacMed Chatbot has demonstrated the potential of simple chatbot technology to support emergency medical care, especially in high-pressure environments like warzones. The success of this tool highlights the need for continued collaboration between medical professionals, NGOs, and technologists to improve healthcare delivery in disaster and conflict settings. In the future, we aim to update the chatbot to incorporate AI to understand questions it doesn't recognize and provide more personalized answers. Future studies can focus on creating chatbots that are adaptable for global use, aiming to build systems that can improve medical education and response during crises around the world. Acknowledgements XR at Yale, Randall Rode, HTWB team

Ahmad Hassan

Marianna Petrea-Imenokhoeva

Stella Nam

+7

Trends in Chronic Kidney Disease–Related Mortality Among Type 2 Diabetes Mellitus Patients in the United States from 1999-2020.

Abstract

Trends in Chronic Kidney Disease–Related Mortality Among Type 2 Diabetes Mellitus Patients in the United States from 1999-2020.

15 June 2025

Background Type 2 Diabetes Mellitus (T2DM) affects around 34.2 million people in the U.S., or 10.5% of the population [1], and is the leading cause of chronic kidney disease (CKD) [2,3]. About 40% of adults with T2DM have some degree of CKD [4]. T2DM has unique mechanisms like insulin resistance and metabolic syndrome that speed up CKD progression. Research [5] has shown that diabetic kidney disease (DKD) in T2DM patients causes albuminuria, declining kidney function, and increased cardiovascular risks. CKD in T2DM patients also raises the risk of heart problems and death. Socio-demographic factors, especially in minority populations, affect outcomes, with non-Hispanic Black and Native American groups facing worse results due to healthcare inequities [6]. Most studies focus on advanced CKD or end-stage renal disease (ESRD), but early-to-moderate stages, where intervention can help, are often overlooked. Objective To determine mortality trends for T2DM-associated CKD from 1999 to 2020 among adults in the United States, emphasizing early-to-moderate CKD stages to highlight opportunities for early intervention. Methods In December 2024, data on CKD-related deaths among patients with T2DM in the United States was collected from the Centers for Disease Control and Prevention’s Wide-Ranging Online Data for Epidemiologic Research database (CDC WONDER) [7]. Death certificates were used to identify CKD and T2DM with ICD-10-CM codes (E11 for T2DM, N18 for CKD). The data was categorized into two age groups: young (25–64 years) and older (>64 years). Institutional review board approval was not needed as the data was de-identified. Data collected included year, population size, demographics, CKD stage, geographic regions, and urban-rural classification. CKD severity was divided into early (stages 1-2), moderate (stages 3-4), and advanced (stage 5 or ESRD). Demographics included age, race/ethnicity, and gender. Urban and rural areas were classified by the National Center for Health Statistics. Crude and age-adjusted mortality rates (AAMRs) per 1,000,000 individuals were calculated. Trends in AAMRs were assessed using Joinpoint regression, and annual percent changes (APCs) were determined with 95% confidence intervals (CIs). Statistical significance was set at p< 0.05. The inclusion criteria included adults aged 25 years and older with CKD listed as a cause of death and T2DM as an underlying or contributing cause. Only deaths occurring in the United States between 1999 and 2020 were included. Additionally, individuals with complete demographic and geographic information in the dataset were considered. The exclusion criteria comprised patients without documented CKD or T2DM on the death certificate, pediatric populations under 25 years, and cases with missing or incomplete data regarding key variables such as age, sex, or geographic classification. Results There were a total of 90,615 CKD-related deaths among adults with T2DM between 1999 and 2020. The AAMR was 15.0 in 1999, which increased to 26.0 by 2009 (APC: 4.20; 95% CI: 1.98-6.48), after which there was a sharp rise to 56.0 by 2012 (APC: 20.5; 95% CI: -0.03-45.28). It then gradually declined to 3.6 (95% CI: 2.8–4.5) in 2015. By 2020, the AAMR steadily rose to 7.5 (APC: 30.5, 95% CI: 18.91-43.2). The AAMRs for men were higher than those for women throughout the years (23.2 with 95% CI of 22.96–23.39 vs.16.1 with 95% CI of 15.97–16.28). NH American Indian or Alaska Native patients had the highest AAMR (56.2, 95% CI: 53.2–59.2]), followed by NH Black or African American (32.0, 95% CI: 31.5–32.6), Hispanic or Latino (28.7, 95% CI: 28.1–29.3), NH Asian or Pacific Islander (16.6, 95% CI: 16.0–17.2), and NH White populations (16.2, 95% CI: 16.1–16.4). Nonmetropolitan areas experienced a relatively greater mortality burden with higher AAMR (23.3, 95% CI: 23.5–24.2) compared to metropolitan areas (16.9, 95% CI: 17.8–18.1), while the Western region had the highest overall AAMR (22.8, 95% CI: 22.5–23.1), followed by the Midwestern (22.1, 95% CI: 21.8–22.4), Southern (18.0, 95% CI: 17.8–18.2), and Northeastern (12.9, 95% CI: 12.7–13.1) regions. Early CKD stages (Stages 1 and 2) showed lower mortality rates, with age-adjusted mortality rates (AAMR) significantly lower than those for later stages. For example, in 2015, the AAMR for individuals in Stage 1 was 3.6 (95% CI: 2.8-4.4), compared to 56.0 (95% CI: 54.2-57.8) for individuals in later stages (Stages 3-5). CKD stages contributed significantly to overall mortality, particularly among minority populations and non-metropolitan areas, underscoring the importance of early detection and management strategies. Discussion Our study found several key trends in CKD and T2DM mortality: First, advanced CKD stages had the highest mortality rates. Mortality increased from 1999 to 2012, declined until 2015, and then rose again from 2015 to 2020. Although the increase in AAMR from 2009 to 2012 was not statistically significant, it may still have practical relevance, as it aligns with broader public health concerns such as rising comorbidities, healthcare disparities, and changes in diagnostic practices. Future research with larger datasets or alternative analytical approaches could further clarify whether this trend represents a true shift in CKD-related mortality among T2DM patients. Men had higher mortality rates than women. Native American and Alaska Native populations had the highest mortality, and non-metropolitan areas had more deaths than metropolitan ones. Mortality rates varied by state, with the Western U.S. showing the highest and the Northeast the lowest. Mortality trends suggest improvements from 2012 to 2015, possibly due to better management of risk factors like hypertension, particularly with the use of SGLT2 inhibitors, which have been shown to improve kidney outcomes in Type 2 diabetes patients [8]. However, mortality increased after 2015, which may be attributed to factors such as changes in healthcare access, socioeconomic disparities, or a deterioration in the management of risk factors and dietary habits. Early CKD stages showed lower mortality, emphasizing the importance of early intervention. Medications like SGLT2 inhibitors, which help slow CKD progression, are key in these stages [8]. We also found racial disparities, with Native Americans and African Americans facing higher mortality rates, consistent with previous studies [9,10]. This is influenced by increased burden of comorbidities like diabetes mellitus, poorly controlled blood pressure, obesity, and liver diseases [9,11,12], social determinants of health, healthcare access, and genetic predispositions as contributors to these disparities. Early CKD stages offer opportunities for intervention to prevent further progression, highlighting the importance of screening, lifestyle changes, and medication. Limitations include reliance on death certificate data and lack of detailed clinical information. Additionally, unmeasured confounders such as socioeconomic status and healthcare access may impact outcomes. Missing data on key variables like medication use may limit the completeness of the analysis. Lastly, findings may not be generalizable to non-U.S. populations with different healthcare practices. Future research should incorporate cohort-based analyses and explore the impact of emerging therapies, such as finerenone and dual GLP-1 receptor agonists, on CKD outcomes in T2DM. Conclusions The highest mortality was seen in men, American Indians, or Alaska Natives, as well as those living in the Western region and non-metropolitan areas. This study highlights the entire spectrum of CKD in T2DM patients, with disparities in mortality burden based on gender, race, and geographic region. Emphasizing early detection and targeted interventions is critical to improving outcomes in this vulnerable population.

Ahmad Hassan

Sana Yameen

Rabia Imtiaz

+3

Demonstration of PinkDetect: A Digital Solution for Breast Health Awareness and Early Detection

Abstract

Demonstration of PinkDetect: A Digital Solution for Breast Health Awareness and Early Detection

25 April 2025

Introduction Breast cancer is a major public health issue in Pakistan, with 1 in 9 women diagnosed in their lifetime and 89% of cases detected at advanced stages. This late detection contributes to approximately 50,000 deaths annually. Key barriers include lack of awareness, social stigma, and limited access to early detection services. PinkDetect addresses these challenges by providing a digital breast health solution that empowers women through education, self-exam tutorials, symptom tracking, and connections to diagnostic services. Technology PinkDetect integrates multiple features to enhance breast health awareness and early detection. It offers culturally sensitive educational content and self-exam tutorials tailored for Pakistani women. A built-in symptom tracker and AI-powered risk assessment tool allow users to monitor their health and receive early evaluations. The app’s diagnostic linkage system, PinkConnect, helps users locate mammography centers and telemedicine services. To ensure data privacy, PinkDetect employs cloud-based storage with encryption. Future advancements include machine learning-driven risk stratification and integration with hospitals for seamless appointment scheduling. Implementation Since its launch in May 2023, PinkDetect has reached over 4,000 women through medical camps, workshops, and focus groups. A pilot program at Jesus and Mary College in India introduced the app to 200 women, while a workshop for visually impaired women in Pakistan emphasized its inclusive approach. Moving forward, PinkDetect aims to expand its reach across Pakistan and South Asia, integrating AI-driven risk assessments and promoting breast cancer awareness at the household level.

Suha Suleman Lalani

Syed Waqas

Solmaz Iranpour

+1

Demonstration of ED Flow Simulation App: Enhancing Emergency Department Efficiency Through Predictive Analytics

Abstract

Demonstration of ED Flow Simulation App: Enhancing Emergency Department Efficiency Through Predictive Analytics

25 April 2025

Introduction Emergency departments (EDs) frequently experience challenges such as overcrowding, long wait times, and inefficient resource allocation, leading to delayed patient care and increased strain on healthcare staff. Addressing these operational inefficiencies requires data-driven solutions that provide real-time insights for hospital administrators. This study introduces the ED Flow Simulation App, a web-based tool designed to optimize patient flow and enhance decision-making in ED settings. Technology The ED Flow Simulation App uses Python for backend processing and Streamlit for frontend visualization, the app provides an interactive and user-friendly experience for hospital administrators. Users can customize key parameters such as patient arrival rates, waiting room capacity, and treatment room capacity to generate real-time visualizations and identify congestion points. The cloud-based storage system ensures data security and enables potential integration with hospital information systems. A live demonstration feature allows users to simulate different hospital scenarios and analyze efficiency metrics. Future enhancements will include machine learning algorithms for more accurate predictions, expanded simulation capabilities for other hospital departments, and an advanced analytics dashboard for deeper insights. Conclusion The ED Flow Simulation App provides a scalable, adaptable solution for managing ED patient flow. With its interactive interface and predictive analytics, it empowers healthcare administrators to make informed decisions, ultimately enhancing efficiency and patient care in emergency settings.

Abdelrahman Mohamed

Olivia Brumfield

Ahmad Hassan

Demonstration of InsureEZ: An AI-Powered Solution to Address Health Insurance Literacy Disparities

Abstract

Demonstration of InsureEZ: An AI-Powered Solution to Address Health Insurance Literacy Disparities

25 April 2025

Health insurance literacy remains a significant barrier for millions of Americans, particularly among vulnerable populations, often leading to poor health outcomes and financial strain. InsureEZ, an AI-driven, web-based platform, aims to address this challenge by simplifying complex health insurance information and empowering users to make informed decisions. Built using Python and large-language models like ChatGPT-4, InsureEZ offers three core features: the Browse Plans Tool for personalized insurance plan comparisons, the Learn About Insurance module for simplified explanations of insurance concepts, and a Resources Page with curated articles and interactive tools. A preliminary study conducted at Rocky Vista University demonstrated the platform’s efficacy, with most participants reporting improved understanding of health insurance and increased confidence in decision-making. InsureEZ’s impact has been recognized through awards at the Rocky Vista University Shark Tank Microgrant Competition and first place at the OMED 24' student poster competition. Future development goals include expanding to mobile platforms, enhancing educational content, and establishing partnerships with healthcare providers and insurers to broaden its reach, especially among underserved populations. By improving accessibility and comprehension of health insurance, InsureEZ has the potential to reduce disparities, improve health outcomes, and promote financial equity nationwide.

Hannah Vedova

Jonathan Scott Wrigley

Sana Khan

+3

Select a journal
Issue