Orris (the CDSS app) is a different tool than what I can directly control, so I can't guarantee its exact output format — but here's a prompt built to get a salient-features-style structured summary from it, along with an evidence-backed differential. **What to upload:** the exit case docx/pptx (or the raw chart photos if it accepts images). **Prompt to paste into Orris:** > Based on the uploaded case (2-year-old female, chief complaint of cough, history of PTB treatment defaulted ~2 months ago, chronic on-and-off cough x1 year worsening x1 week, fever, subcostal retractions, vomiting with phlegm, family history of bronchial asthma, RR 53, HR 153, crackles at both lung bases, wheezing), please generate: > > 1. **Salient Features** — organized as a table with two columns: Pertinent Positives and Pertinent Negatives, grouped by category (General Data/History, HPI, Past Medical History, Family History, Physical Exam). > 2. **Working Impression** — most likely diagnosis with 1–2 differentials to rule out, each with a one-line rationale. > 3. **Severity/Risk Classification** — using the current Philippine PAPP/PIDSP PCAP CPG criteria (or applicable pediatric pneumonia severity guideline) to justify high-risk vs. low-risk classification. > 4. **Evidence-backed diagnostic work-up** — with your source citations, prioritized by what would confirm/exclude each item in the working impression. > > Please cite your sources for each recommendation, and flag anything in the case data that's ambiguous or would need clinical confirmation rather than assuming it. A couple of practical notes: - I built in the "flag ambiguous data" instruction since your chart had genuinely illegible parts (O2 sat, ear exam, etc.) — worth keeping that guardrail in any tool you use on this case. - Since Orris pulls from its own literature search rather than the exact PIDSP 2021 CPG document I fetched directly, double-check its cited sources/doses against the actual guideline before trusting them for your defense — CDSS tools can cite plausible-sounding but wrong specifics.