From hazard ratios to trial design — statistics in the language of cancer medicine.
This programme turns the statistics behind modern oncology into working knowledge. Across a year, practising and trainee oncologists learn to read a trial correctly, judge why a design was chosen, and run the core analyses themselves — with the syllabus deliberately weighted toward the two things you meet every week: time-to-event analysis and clinical trial design. Every session ends with a hands-on task, so the methods are practised, not just heard.
The survival and trial-design blocks are the core of the syllabus; foundations can be compressed for an experienced cohort.
| Block | Title | Sessions | Tasks |
|---|---|---|---|
| 01 | Foundations in Cancer Data | 1 – 4 | 8 |
| 02 | Comparing Groups | 5 – 7 | 6 |
| 03 | Survival Analysis Core | 8 – 13 | 18 |
| 04 | Oncology Trial Design Core | 14 – 19 | 12 |
| 05 | Biomarkers, Synthesis & Real-World Data | 20 – 24 | 10 |
| 06 | Appraisal & Capstone | 25 – 26 | 2 + project |
The groundwork, framed in oncology terms.
The oncology evidence hierarchy; observational versus experimental designs; and how bias, confounding and immortal-time bias arise in cancer research.
Clinical, pathology, registry and genomic data types; building an analysis-ready dataset; and constructing a clear baseline "Table 1".
Sensitivity, specificity, PPV/NPV and likelihood ratios; ROC curves for tumour markers; and lead-time and length-time bias in screening.
Risk ratio, odds ratio, hazard ratio and number-needed-to-treat; absolute versus relative benefit for patient communication; reading confidence intervals.
Or register for the entire year and save.
Register for full year — ₹2,000The inference toolkit.
Type I and II error; multiplicity across multiple endpoints and interim looks; one- versus two-sided testing; and why a p-value is never the whole story.
t-tests and ANOVA; chi-square and Fisher's exact tests for objective response rates; paired comparisons; and choosing the right test.
Wilcoxon and Kruskal–Wallis tests; analysing ordinal RECIST-category data; and reading and building waterfall plots.
Or register for the entire year and save.
Register for full year — ₹2,000The heart of oncology statistics.
Censoring; defining OS, PFS and DFS precisely; Kaplan–Meier estimation; and median and landmark survival probabilities.
The log-rank and stratified log-rank tests; number-at-risk tables; and the pitfalls of crossing curves and early-versus-late separation.
Hazard ratios and what they really mean; covariate adjustment and stratification; and checking the proportional-hazards assumption.
Time-varying covariates; non-proportional hazards under immunotherapy; and restricted mean survival time as an interpretable alternative.
Cause-specific versus subdistribution hazards; cumulative incidence of cancer death versus other-cause death; and the Fine–Gray model.
Illness-death and progression-to-death models that capture the whole disease course rather than a single endpoint.
The core of the syllabus. Or register for the entire year and save.
Register for full year — ₹2,000Phase I through platform trials.
OS versus PFS versus ORR versus duration of response; surrogate endpoints and their validation; and the estimands framework for intercurrent events.
Toxicity and the MTD; rule-based 3+3 versus model-based CRM and BOIN designs; and dose escalation and expansion cohorts.
Simon's two-stage design; single-arm versus randomised phase II; go/no-go decision rules; and futility stopping.
Randomisation, stratification and blinding; intention-to-treat versus per-protocol; analysis populations; and the logic of event-driven trials.
Powering survival endpoints by number of events; hazard-ratio-based sample size; and non-inferiority and equivalence designs.
Interim analyses and stopping boundaries; basket and umbrella trials; master protocols; and Bayesian adaptive randomisation.
Or register for the entire year and save.
Register for full year — ₹2,000Evidence beyond the single trial.
The crucial distinction; subgroup-analysis pitfalls; interaction tests; multiplicity across subgroups; and biomarker-stratified designs.
Logistic regression for response; Poisson and negative-binomial models for adverse-event rates; and building and validating a prognostic model.
Pooling hazard ratios; fixed versus random effects; forest and funnel plots; heterogeneity; and an introduction to network meta-analysis.
Registry and EHR data; propensity scores and matching; external and synthetic control arms; and target-trial emulation.
Informative censoring; missingness in patient-reported and QoL data; multiple imputation; and sensitivity analyses.
Or register for the entire year and save.
Register for full year — ₹2,000Putting it all together.
Reading an oncology trial paper critically; CONSORT and RECIST reporting; interpreting Kaplan–Meier curves and forest plots at a glance; and spotting common statistical errors.
Participants take an oncology dataset or a published trial end-to-end — design critique, analysis and interpretation — and present for discussion and feedback.
Or register for the entire year and save.
Register for full year — ₹2,000Run entirely online over live video and recorded for catch-up — no travel, with scheduling that accommodates participants across time zones.
56 guided tasks plus the capstone — most sessions carry two, the survival block carries three, and every task uses real or realistic oncology data.
Each session is anchored in a familiar landmark trial, keeping the statistics concrete and clinically motivated.
One or two real oncology datasets run across the year, so participants see the same patients analysed many different ways.
The survival and trial-design blocks are the core; foundations can be compressed for an experienced cohort.
Demonstrations are in R; the statistical content maps directly onto SPSS or Stata for those who prefer them.
Reserve your seat for the full year, or start with a single session that interests you. Registration takes a minute; payment is via Razorpay UPI.