Blog

1106 posts on clinical medicine, epidemiology, research methodology, statistics and code — in English and Thai. Filter by category and language, or search.

Learning path · เส้นทางการเรียนรู้

For medical studentsสำหรับนักศึกษาแพทย์

Build clinical understanding, interpret tests, and learn to read evidence one step at a time.

ปูพื้นฐานทางคลินิก แปลผลการตรวจ และฝึกอ่านหลักฐานทีละขั้น

Start learning · เริ่มเรียนรู้ →
Learning path · เส้นทางการเรียนรู้

For doctors (MD)สำหรับแพทย์

Start with a clinical question. Appraise the evidence, interpret diagnostic tests, and examine how prediction models apply to practice.

เริ่มจากคำถามในเวชปฏิบัติ ประเมินหลักฐาน แปลผลการตรวจวินิจฉัย และพิจารณาการใช้แบบจำลองพยากรณ์ในบริบททางคลินิก

Start learning · เริ่มเรียนรู้ →
Learning path · เส้นทางการเรียนรู้

For researchersสำหรับนักวิจัย

Follow a question from study design through analysis, prediction and evidence synthesis.

พัฒนาคำถามวิจัยตั้งแต่การออกแบบ การวิเคราะห์ การพยากรณ์ ไปจนถึงการสังเคราะห์หลักฐาน

Start learning · เริ่มเรียนรู้ →
Interaction: At What Reference Level, on What Scale?
Clinical Epidemiology ResearchUniqcret doctor knowledgesMethodology and Research Design

Interaction: At What Reference Level, on What Scale?

With an interaction term, a main effect is read at the reference level of the other variable, and an interaction finding holds only on a named scale: the same four risks can show none on one scale and some on another.

ENupdated 8 Oct 2026· 17 min read
Same Cohort, Two Odds Ratios: Non-Collapsibility Without Confounding
Clinical Epidemiology ResearchUniqcret doctor knowledgesData Analytics or Statistics

Same Cohort, Two Odds Ratios: Non-Collapsibility Without Confounding

One cohort, an adjusted regression and a weighted analysis, both done correctly, and two different odds ratios: the two methods may be answering different questions.

ENupdated 8 Oct 2026· 10 min read
Getting a Risk Ratio for a Common Outcome: The Step-Down When Log-Binomial Hits the Wall
Clinical Epidemiology ResearchUniqcret doctor knowledgesData Analytics or Statistics

Getting a Risk Ratio for a Common Outcome: The Step-Down When Log-Binomial Hits the Wall

A binary outcome, a risk ratio to report and a log-binomial model that will not converge: step down to modified Poisson with a robust SE, and see why the log link gives the risk ratio and where the wall at a risk of 1 comes from.

ENupdated 8 Oct 2026· 29 min read
Kaplan-Meier and Cox for a First Event: KM Describes, Cox Models
Clinical Epidemiology ResearchUniqcret doctor knowledgesMethodology and Research Design

Kaplan-Meier and Cox for a First Event: KM Describes, Cox Models

For the time to a first event, Kaplan-Meier describes the probability of staying event-free and the Cox model estimates the hazard ratio, adjusted if needed, as long as that ratio stays roughly constant.

ENupdated 8 Oct 2026· 16 min read
IPTW: Making Treatment Groups Comparable and Checking That It Worked
Clinical Epidemiology ResearchUniqcret doctor knowledgesMethodology and Research Design

IPTW: Making Treatment Groups Comparable and Checking That It Worked

Weighting changes how much each patient counts, not who is in the data. Here is how IPTW makes treatment groups comparable on measured characteristics, and the three checks that show it worked.

ENupdated 8 Oct 2026· 12 min read
Blog — Uniqcret