← All work samples

AI Evaluation / Research Synthesis / Data Quality Work Sample

Medical Record Evidence Synthesis

ENT referral timeline, findings & evidence gaps

Aligned a referral, medication list, imaging report and laboratory data; separated confirmed findings from suspected conditions; and preserved missing-evidence boundaries.

Scope

4unique record or deliverable types
3supporting formats: DOCX, PDF and CSV
24CBC/CMP analyte rows in the surviving lab file
Multipleduplicate copies identified and normalized

01

Overview

This simulated ENT-focused assignment involved a smaller but different kind of quality problem: turning records created for different purposes into a concise referral narrative without converting suggestive evidence into a confirmed diagnosis. The surviving file set includes a completed referral deliverable and three supporting record types, plus duplicate exports.

02

Challenge

The records spanned different dates and purposes. A brain MRI contained only incidental sinus information, the medication list warned that outside changes might be missing, and the referral referenced a dedicated sinus CT whose report was not preserved in the available portfolio source set.

03

Source material

  • Active medication list generated 12/04/2025
  • Brain MRI report from 08/27/2025
  • CBC/CMP laboratory export from 03/18/2026
  • PCP-to-ENT referral letter dated 02/28/2026
  • Duplicate DOCX, PDF and CSV exports of the same underlying records

04

My work

  • Matched patient identity across the surviving records and removed duplicate copies from the scope count.
  • Reconstructed the sequence of treatment history, evolving symptoms, office examination and referral request.
  • Separated what the imaging established from incidental findings and from the clinician's suspected condition.
  • Preserved the distinction between a rigid nasal examination and the unperformed laryngeal assessment.
  • Kept later laboratory abnormalities in their proper time context and did not use them as proof of causation.
  • Documented the missing CT report as an evidence boundary rather than inventing CT findings.

05

Analytical approach

01

Normalize the record setIdentify duplicate exports and count unique information sources only once.

02

Build the chronologyOrder prior imaging, medication history, treatment attempts, new symptoms, examination and later labs by date.

03

Label evidence typeSeparate reported symptoms, observed examination findings, imaging findings, treatment history and proposed next steps.

04

Distinguish fact from suspicionDo not turn concern for chronic rhinosinusitis or a possible laryngeal component into a confirmed diagnosis.

05

Name what is missingLimit the synthesis when a referenced report is not present in the available source set.

06

Key findings

Specific examples from the completed work.

Imaging scope kept accurate

The MRI was a brain study for headaches. It found no acute intracranial abnormality and only incidental mild bilateral maxillary mucosal thickening; it was not treated as a dedicated sinus evaluation.

Treatment history reconciled

The medication and referral records supported prior intranasal steroid, antihistamine, saline and antibiotic use with incomplete or transient improvement, while also warning that the older primary-care medication list might not capture later changes.

Observed versus unassessed anatomy

The referral documented right middle-meatus discharge, edematous turbinates and a rightward septal deflection. It also explicitly stated that the nasopharynx and larynx were not assessed on that rigid examination.

Later labs not overinterpreted

The 03/18 laboratory file was mostly within its listed ranges, with eosinophils 8.4% and absolute eosinophils 596 flagged high. Because the labs post-dated the 02/28 referral, the portfolio does not present them as the cause of earlier symptoms.

Missing evidence preserved

The referral says a dedicated sinus CT was completed on 03/14, but that report is not in the surviving source set. No CT-specific finding is attributed in this case study.

07

Deliverable

A PCP referral letter that consolidated symptoms, prior treatment, observed sinonasal findings, limits of the office examination, and the requested specialist evaluation without asserting a final diagnosis. This portfolio adaptation adds a clear source-scope and evidence-gap note.

08

Skills demonstrated

  • Medical-record synthesis
  • Timeline reconstruction
  • Duplicate normalization
  • Fact-versus-inference separation
  • Evidence-gap detection
  • Cross-file identity matching
  • Plain-language research summary

Why it matters

This case shows a different evaluation strength: concise synthesis under uncertainty. I can combine records written for different purposes while preserving chronology, source limitations and the difference between established findings and plausible next questions.