Amir Erfan Zareei Shams Abadi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Amir Erfan Zareei Shams Abadi has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Dima Dandachi
Division of Infectious Diseases, University of Missouri, Columbia, United States; Medical Director, Columbia/Boone County Public Health & Human Services, Columbia, USA, USA
1 shared publication - Iris Zachary
Missouri Cancer Registry and Research Center, Department of Public Health, University of Missouri, Columbia, USA
1 shared publication - Karan Karthik
Missouri Cancer Registry and Research Center, University of Missouri, Columbia, MO 65211, USA
1 shared publication - Maryam Seifaddini
Missouri Cancer Registry and Research Center, University of Missouri, Columbia, MO 65211, USA
1 shared publication - Mirna Becevic
Department of Dermatology; Missouri Telehealth Network; Institute for Data Science and Informatics, University of Missouri, Columbia, MO, USA
1 shared publication - Mohammad Beheshti
Missouri Cancer Registry and Research Center, Department of Public Health, University of Missouri, Columbia, USA
1 shared publication
Latest Publications
- Open Access
- 18 September 2026
- Published by MDPI AG in Information
- Vol. 17 (9) , 917
Entity Resolution Using Transformer-Based Language Models: A Systematic Scoping Review
Entity resolution (ER) is fundamental to integrating heterogeneous data, which traditional approaches address through deterministic rule-based methods and probabilistic record linkage. We conducted a systematic scoping review following PRISMA-ScR to characterize the use of transformer-based language models for ER. Five databases were searched, and 155 studies were included for synthesis. The literature expanded sharply after 2023, with 55% of included studies published in 2025 or 2026. General-purpose matching was the most common entity focus, followed by product/e-commerce. Healthcare applications were especially scarce, with only one study applying ER to the patient/healthcare domain. Among studies that performed blocking, embedding-based similarity was most common, followed by string-based approaches. Classification-head approaches remained the most common matching approach, followed by prompt-based approaches. Encoder-only models remained the most widely evaluated architecture, while decoder-only models grew increasingly prominent from 2023 onward. Full-parameter fine-tuning was the predominant learning strategy, followed by zero-shot and few-shot prompting. Among studies classified as general-purpose, nearly one-third were evaluated on only one or two entity types, limiting their generalizability. Reported best F1 scores varied across benchmarks, with no consistent advantage for encoder-only versus decoder-only architectures. These findings support the need for more diverse, end-to-end evaluations that consider efficiency and robustness alongside accuracy.
- 18 September 2026
- Published by MDPI AG in Information
- Vol. 17 (9) , 917
Open AccessEntity Resolution Using Transformer-Based Language Models: A Systematic Scoping Review
Entity resolution (ER) is fundamental to integrating heterogeneous data, which traditional approaches address through deterministic rule-based methods and probabilistic record linkage. We conducted a systematic scoping review following PRISMA-ScR to characterize the use of transformer-based language models for ER. Five databases were searched, and 155 studies were included for synthesis. The literature expanded sharply after 2023, with 55% of included studies published in 2025 or 2026. General-purpose matching was the most common entity focus, followed by product/e-commerce. Healthcare applications were especially scarce, with only one study applying ER to the patient/healthcare domain. Among studies that performed blocking, embedding-based similarity was most common, followed by string-based approaches. Classification-head approaches remained the most common matching approach, followed by prompt-based approaches. Encoder-only models remained the most widely evaluated architecture, while decoder-only models grew increasingly prominent from 2023 onward. Full-parameter fine-tuning was the predominant learning strategy, followed by zero-shot and few-shot prompting. Among studies classified as general-purpose, nearly one-third were evaluated on only one or two entity types, limiting their generalizability. Reported best F1 scores varied across benchmarks, with no consistent advantage for encoder-only versus decoder-only architectures. These findings support the need for more diverse, end-to-end evaluations that consider efficiency and robustness alongside accuracy.
- Open Access
LLM-Assisted Taxonomy and Temporal Analysis of Provider Questions About HIV in provider-to-provider telehealth
Introduction: Ongoing education in HIV care is limited for many healthcare providers working in rural and non-academic settings, which can reduce patients’ access to high-quality care. To guide targeted tele-mentoring and continuing education, we analyzed questions submitted by clinicians during Extension for Community Healthcare Outcomes (ECHO) sessions to characterize learning needs and thematic trends. Methods: We reviewed 78 clinical questions submitted during Project ECHO sessions and developed a structured classification of topics raised by clinicians. Using text-embedding representations and large language models (LLMs), we explored automated approaches to categorize questions and identify thematic clusters. Analyses compared the distribution of topics across professional roles to detect role-specific learning needs. Results: Distinct topic patterns emerged by clinician type. Physicians and pharmacists most often asked about initiating and modifying antiretroviral therapy (ART). Nurse practitioners focused on ART and adherence support, while allied health professionals and PAs raised social-support and care-navigation issues. Medication-related questions frequently highlighted adherence concerns and ART change considerations. Discussion: ECHO questions reveal clear, role-dependent learning needs that can inform targeted tele-mentoring. LLM-based embeddings provided a practical, scalable way to classify questions and monitor trends, supporting more tailored HIV training for different provider groups.
Open AccessLLM-Assisted Taxonomy and Temporal Analysis of Provider Questions About HIV in provider-to-provider telehealth
Introduction: Ongoing education in HIV care is limited for many healthcare providers working in rural and non-academic settings, which can reduce patients’ access to high-quality care. To guide targeted tele-mentoring and continuing education, we analyzed questions submitted by clinicians during Extension for Community Healthcare Outcomes (ECHO) sessions to characterize learning needs and thematic trends. Methods: We reviewed 78 clinical questions submitted during Project ECHO sessions and developed a structured classification of topics raised by clinicians. Using text-embedding representations and large language models (LLMs), we explored automated approaches to categorize questions and identify thematic clusters. Analyses compared the distribution of topics across professional roles to detect role-specific learning needs. Results: Distinct topic patterns emerged by clinician type. Physicians and pharmacists most often asked about initiating and modifying antiretroviral therapy (ART). Nurse practitioners focused on ART and adherence support, while allied health professionals and PAs raised social-support and care-navigation issues. Medication-related questions frequently highlighted adherence concerns and ART change considerations. Discussion: ECHO questions reveal clear, role-dependent learning needs that can inform targeted tele-mentoring. LLM-based embeddings provided a practical, scalable way to classify questions and monitor trends, supporting more tailored HIV training for different provider groups.