Use of Machine Learning to Derive Discrete Clinical Phenotypes and Assess Treatment-Effect Heterogeneity in the Steroids in Cardiac Surgery Trial Dataset
Bibliographic record
Abstract
BACKGROUND: Robust clinical trial data provide a key component for the development of evidence-informed medicine. However, clinical trial data may demonstrate treatment-effect heterogeneity, where some patients benefit from an intervention while others receive no benefit or perhaps even harm. If so, targeted therapy or a "personalized medicine" approach could provide treatment to a certain patient subset, that is, a specific clinical phenotype, who are most likely to benefit. Using data from the Steroids in Cardiac Surgery (SIRS) clinical trial, we tested the hypothesis that methylprednisolone, which did not have a significant effect on mortality or major morbidity, improves outcomes in 1 or more clinical phenotypes. METHODS: We used the partitioning around medoids algorithm to derive phenotypic clusters using 30 preoperative variables in a Cleveland Clinic cardiac surgery developmental dataset. Patients in the SIRS trial were assigned to the derived clusters. Methylprednisolone-response heterogeneity was evaluated among SIRS patients for the coprimary outcomes of 30-day mortality and composite of death and major morbidity. This was accomplished by fitting separate logistic regression models for each outcome and evaluating the interaction between treatment groups and assigned phenotypic cluster. RESULTS: The 16,395 patients in the developmental dataset were clustered into 4 phenotypes: younger and healthier; mid-age and moderately sick; oldest, sicker and more aortic valve surgery; sickest, more coronary artery bypass grafting (CABG) and low left ventricular ejection fraction (LVEF). The phenotypes had differing risk profiles and were associated with patient outcomes. For example, patients in sickest, high CABG and low LVEF group were at highest risk amongst all phenotypes, with significantly increased odds of experiencing a composite of mortality and severe morbidity (odds ratio [OR]: 3.4, 95% confidence interval [CI], 2.4-4.8) compared to younger and healthier group. When clustering was applied to the SIRS trial dataset (N = 6836), patients in the sickest, more CABG and low LVEF group similarly represented the highest-risk category for mortality and severe morbidity (OR: 2.1; 95% CI, 1.6-2.9). After examining the treatment-effect in each phenotype, we did not find evidence that methylprednisolone treatment-effect on the coprimary outcomes differed by phenotypes (all treatment-phenotype interaction term P > .05). CONCLUSIONS: Despite substantial differences in preoperative risk profiles, findings were neutral for methylprednisolone across all phenotypes. However, the general concept of evaluating trial results across clinical phenotypes represents a novel approach to identify subgroup differences in treatment-effect and to collect preliminary evidence of potential benefit with targeted therapy.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.026 | 0.061 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.003 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.000 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".