Best prompts for LIM for summarization using medical research

You are given a full medical research article (or sections). Produce a structured 250‑word summary with the following labeled sections: Background, Objective, Methods (design, population, key interventions/comparators, primary outcome), Results (numerical effect sizes with measures of uncertainty and sample sizes), Conclusion, and One‑sentence clinical takeaway. Preserve all numeric values and units exactly as in the source. If an article identifier (PMID/DOI) is provided, include it on the final line. Output as plain text with those headings only. Input: [PASTE_FULL_PAPER_OR_ABSTRACT_HERE]

Given a clinical trial or comparative study text, extract and return a PICO table (Population, Intervention, Comparator, Outcomes) and for each primary outcome provide: absolute event rates, relative effect (RR/OR/HR as reported), 95% CI, p‑value, sample sizes per arm, and a one‑line interpretation of clinical impact. Output as a numbered list with labeled fields. Input: [PASTE_ABSTRACT_OR_RESULTS_SECTION_HERE]

Create a plain‑language summary of the provided study suitable for patients or caregivers, maximum 150 words. Include: why the study was done, what was compared, main findings with simple numbers (e.g., "4 of 100"), what it means for patients, and any important safety points. Avoid jargon; explain any necessary medical terms in one short parenthesis. Input: [PASTE_ABSTRACT_OR_RESULTS_SECTION_HERE]

You will receive an image of a figure (e.g., Kaplan‑Meier curve, forest plot, or bar chart) plus its caption. Describe in 6–8 bullet points: (1) figure type, (2) population/arms shown, (3) main quantitative findings (hazard ratios, CIs, p‑values, point estimates), (4) direction and magnitude of effect, (5) any visually apparent subgroup differences, and (6) one caveat based on axes/scale/labels. Keep bullets short and include the exact numeric labels present in the image. Input: [UPLOAD_FIGURE_IMAGE] Caption: [PASTE_CAPTION_HERE]

Given the methods and results sections, produce 6–10 concise bulleted items identifying risk of bias domains (randomization, allocation concealment, blinding, outcome reporting, attrition, selective reporting, conflicts of interest), each bullet containing: domain label, brief evidence from the text, and an overall judgement (Low/Some concerns/High). End with two short suggestions to reduce each identified high‑risk concern in future studies. Input: [PASTE_METHODS_AND_RESULTS_SECTIONS_HERE]

Provide a clinician‑oriented explanation of the study's statistical findings. For each reported metric (mean difference, OR, RR, HR, p‑value, CI), explain in one sentence what the number means clinically, comment on precision and clinical importance, and note any multiple testing or subgroup caveats. Keep to 5–7 lines maximum. Input: [PASTE_RESULTS_TABLE_OR_TEXT_HERE]