SmartMedCode supports your professional judgement. You confirm every final code.
Portfolio
Ugwu Tochi Peace
Licensed Health Information Management professional with more than 15 years of hospital experience in Nigeria. Founder of HIMForward.
No certification body, degree, client, award or employer is named here. The licence body and number are not published here.
- Licensed Health Information Management professional. The licence body and number are not published here.
- More than 15 years of hospital experience in Nigeria.
- Founder of HIMForward.
Email: contact@himforward.com
Sample case studies
Results below are calculated from the sample files in this app. They are not operational hospital results.
1. Medical coding quality validation
Problem. Assigned codes need a repeatable screen before a person decides what to correct.
Method. The validator runs the Stage 1 rules. The audit file then compares assigned mains with a prepared reference.
Results. On the sample audit file, 11 of 20 main codes match. Accuracy 55.0%. Error rate 45.0%. Rows needing correction under the answer key: 12.
Limitations. The rules are not a full ICD-11 rulebook. WHO was not required for this count.
What it establishes. The author can define a screen, separate warnings from confirmed errors, and report a metric from rows rather than a fixed percentage.
2. Healthcare data quality assessment
Problem. A coding file can look finished while ages, sexes or identifiers are unfit to use.
Method. 38 sample encounters are filtered by department, coder, classification and encounter type. Completeness and screening findings are recalculated from the visible rows.
Results. 37 of 38 encounters have an age in years or days. Other rates are on the dashboard so they move with the filter instead of being copied here as a slogan.
Limitations. The benchmark slider is a sample target, not an official standard. The page does not produce NHMIS or DHIS2 files.
What it establishes. The author can turn a row set into a filtered quality view and an export.
3. Healthcare AI annotation review
Problem. Suggested codes need a human status, a reason, and an export a supervisor can read.
Method. 14 sample phrases were paired by hand with published MMS titles. Reviewers set status and equivalence. Two pairs are intentionally poor.
Results. There is no model accuracy figure. Confidence is empty on every row because no model ran.
Limitations. The pairs are not a WHO map and not an ICD-10 crosswalk.
What it establishes. The author can review terminology suggestions without dressing a manual sample up as machine learning.
Services
Medical coding quality review
A record-level worksheet of screening findings, with rule, severity, and a note that a coder must verify before any code changes.
Contact meHealthcare data validation
A check of required fields, formats and internal duplicates on a de-identified file, plus a short exception list.
Contact meCoding audit support
A comparison of assigned codes to an agreed reference, with main-condition accuracy, error rate and discrepancy classes counted from the file.
Contact meHealthcare data quality reporting
A filtered summary of completeness, unspecified-code frequency and department patterns. Not an official NHMIS or DHIS2 submission.
Contact meAI annotation quality review
A review sheet for suggested codes: status, equivalence, reason and explanation, exported as CSV or JSON.
Contact meTerminology mapping review
A human review of proposed targets against source phrases. Official maps are used only when a verified WHO table is supplied.
Contact meContact
Email contact@himforward.com. This page does not claim that paid contracts have already been completed.
Contact Me