1 Vacancy

Job Description

Project 1: To create individual risk matrix between long-term drug and cancer

Cancer has increasingly been recognized as a chronic disease, with its treatment accounting for 13% of Taiwan's total health insurance expenses. Effective cancer treatment planning and benefit assessments should include preventive strategies beyond traditional therapeutic approaches. Chemoprevention, the use of medications to reduce cancer risk, is a well-established concept. Epidemiological studies have shown that long-term use of certain chronic disease medications is significantly linked to a lower risk of specific cancers. To analyze the relationship between chronic medication use and cancer risk, we must consider factors such as drug types, cancer categories, age, and gender, utilizing multiple analytical approaches. This complex issue cannot be addressed with a single method, but it requires leveraging medical informatics. By utilizing big medical data and applying data mining techniques, we can systematically assess the associations between the long-term use of chronic medications and cancer risk. Visualization techniques will help make these insights accessible, enabling the development of personalized risk matrices.

This short-term internship aims to leverage big medical data and data mining techniques to systematically evaluate how various drug types, patient age, gender, and cancer categories interact.

Key Responsibilities

1. Literature Review: Conduct thorough searches (e.g., PubMed, Google Scholar) to gather existing research on chemoprevention, chronic medication, and cancer risk. Synthesize findings into summaries or annotated bibliographies for reference by the research team.
2. Data Compilation & Preparation: Assist in collecting, cleaning, and organizing relevant datasets (e.g., patient demographics, medication records, cancer incidence).
3. Report Preparation & Visualization: Draft written reports outlining key insights, trends, and preliminary conclusions on the associations between medication use and cancer risk.

Relevant Publications:

1. https://doi.org/10.1111/cas.16422
2. https://doi.org/10.3390/ijms24043814
3. https://doi.org/10.3390/cancers14246083

 

Project 2: Utilizing Large Language Models and Real-World Data to Enhance Medication Safety: Creating RxLlama

The emergence of large language models (LLMs), such as ChatGPT, offers new approaches to improving medication safety due to their powerful data processing, interactive learning, and dialogue capabilities. However, these models largely depend on Internet data and lack access to critical resources like electronic medical records and expert evaluations. Moreover, using cloud-based LLMs poses risks to patient data security, raising concerns for clinical use. This project, titled "Utilizing Large Language Models and Real-World Data to Enhance Medication Safety: Creating RxLlama," aims to leverage the capabilities of open-source LLMs (specifically Llama) to improve prescription safety through advanced prompt engineering and real-world data integration. It will involve model refinement, validation in medical settings, and exploration of educational and clinical applications. By integrating human feedback and external medical databases, we seek to align the model with expert knowledge for continuous improvement.

Join a short-term research project titled "Utilizing Large Language Models (LLMs) and Real-World Data to Enhance Medication Safety: Creating RxLlama." You will assist in integrating open-source LLMs (e.g., Llama) with clinical data to improve prescribing practices and reduce medication errors.

Key Responsibilities

1. Data Preparation: Collect and preprocess sample clinical or pharmacy-related datasets.
2. Prompt Engineering & Model Testing: Experiment with prompt designs and test LLM outputs for accuracy and clarity in medication safety contexts.
3. Reporting & Documentation: Summarize findings, create visualizations, and present suggestions to the project team for model refinement.

Relevant Publications:

1. https://doi.org/10.1016/j.cmpb.2021.106181
2. https://doi.org/10.1016/j.hlpt.2024.100852

 

Preferred Intern Education Level

Master’s and Ph.D. Candidates

• Fields: Healthcare Informatics, Biomedical Informatics, Public Health, Pharmacy, Computer Science, Data Science, Machine Learning, Biostatistics, or other Health/Life Science disciplines.
• Reasoning: Graduate-level students typically bring deeper research experience, methodological rigor, and specialized skills suited for data analysis, model development, and interdisciplinary collaboration.

Recent or Current Ph.D. Students

• Fields: Computer Science, Artificial Intelligence, Machine Learning, Data Science, Biostatistics, or related domains with a strong healthcare/medical focus.
• Reasoning: Doctoral candidates or recent graduates offer advanced research capabilities, leadership potential in method design, and the ability to tackle complex analytical challenges—especially those involving large datasets and specialized methodologies.

Skill sets or Qualities

Healthcare Data Familiarity

• Basic understanding of electronic medical record (EMR) structures and their practical uses in research

Data Analysis

• Proficiency in SAS, Python, or R for data manipulation, cleaning, and exploratory analysis
• Comfort with common data analysis workflows, including simple descriptive statistics and data visualization

Machine Learning

• Foundational understanding of machine learning concepts (e.g., supervised/unsupervised learning, model evaluation)
• Ability to apply basic algorithms or models under guidance

Communication & Collaboration

• Strong written and verbal communication skills in English (additional language skills are a bonus)
• Comfortable presenting technical findings to non-technical audiences, such as clinicians or policymakers
• Team-oriented mindset, open to feedback, and eager to collaborate with researchers from diverse backgrounds