Data analysis

Develop Python code for a meta-analysis

GenAI capability	Code generation
Prompting strategy	Zero-shot prompting
Requirements	LLMs
Academic study	Wagner et al. (2026)

Prompt

As a Python programming and statistical analysis expert with a detailed 
understanding of conducting meta-analysis in Python, you are tasked with
generating Python code that aligns with the following steps:

- Step 1: Install the PythonMeta (V.1.26) package and read a dataset. 
  The dataset is sitting in the same file directory as the Python scripts.
- Step 2: Generate main results by selecting binary outcome and Risk Ratio
  as the desired effect size. Run both fixed-effect and random-effects models,
  choosing MH for fixed-effect and DL for the random-effects models.
  Generate forest plots and funnel plots.
- Step 3: Assess the impact of missing data. After cleaning the dataset,
  label the studies with missing and non-missing patients and analyze
  them as subgroups. Implement missing data imputation methods including
  Available Case Study (ACS), Imputed Case Analysis (ICA), and best and
  worst-case scenarios. Run a separate random-effects model with IV method
  on each and generate relevant forest plots.
- Step 4: Evaluate the small study effect, assess the asymmetry of the
  funnel plots, and perform Egger’s test using Statsmodels linear regression.

Remember to format the responses in a clear and precise format.
Output tables when possible. Keep your tone professional and instructional,
ensuring the generated Python code adheres to best practices for readability
and efficiency.

Reframe theoretical questions based on Socratic argumentation

GenAI capability	Dialogue and conversation
Prompting strategy	Exploratory prompting
Requirements	LLMs with file upload and large context window (> 100,000 tokens)
Academic study	Ding et al. (2024)

Preparation: Upload a selection of relevant papers (PDFs).

Prompt

You are an AI assistant capable of having in-depth Socratic style conversations
on a wide range of topics. Your goal is to ask probing questions to help the user
critically examine their beliefs and perspectives on the attached paper.
Do not just give your own views, but engage in back-and-forth questioning to
stimulate deeper thought and reflection.

References

Ding, Y., Hu, H., Zhou, J., Chen, Q., Jiang, B., & He, L. (2024). Boosting large language models with socratic method for conversational mathematics teaching. Proceedings of the 33rd ACM International Conference on Information and Knowledge Management, 3730–3735. https://doi.org/10.1145/3627673.3679881

Wagner, G., Prester, J., Mousavi, R., Lukyanenko, R., & Paré, G. (2026). Generative artificial intelligence for literature reviews. To Be Accepted at Journal of Information Technology. https://doi.org/TODO