Cluster 2: improve · prediction · tumor
Automatically detected text-similarity community. Keywords are calculated from article titles and summaries; membership is recalculated when the published corpus changes.
Do two immune-biology scores improve response prediction across urothelial cancer cohorts?
Fixed Teff and stromal features evaluated across public urothelial cancer cohorts, with discrimination, calibration, missingness and transport limits.
- Adding fixed T-effector and stromal RNA scores to a minimal clinical model raised cross-validated response AUC from 0.509 to 0.645 in the development cohort (298 patients, 68 responders).
- In the external-source cohort (72 patients, 11 responders), the fixed model reached AUC 0.554 (95% interval 0.367–0.725) and probability calibration was poor.
- The two scores support a biological hypothesis worth further study, not a clinically useful biomarker, an individual response probability or a prediction of who benefits from checkpoint inhibition.
What changes in paired tumor immune-expression measurements during nivolumab?
Fixed immune scores in 43 public RNA pairs, with paired change, normalization sensitivity and patient-separated response-association diagnostics.
- In 43 public pre/on-nivolumab RNA pairs, a fixed cytolytic score rose in 33 of 43 participants (median +0.346, 95% interval +0.154 to +1.108), stable across FPKM and CPM (Spearman 0.996).
- Adding the change to baseline expression raised held-out response AUC from 0.599 to 0.673, but the interval (−0.084 to +0.269) included no improvement, with only nine responders.
- An observed on-treatment change is pharmacodynamic evidence, not a qualified predictive biomarker. Biopsy timing and selection into a second biopsy limit its interpretation at a fixed information time.
Does separating PSA decline and regrowth improve early prediction?
A public clinical PSA dataset used to compare landmark prediction, uncertainty calibration and decline/regrowth identifiability against simple benchmarks.
- In 400 patients from a public PSA dataset, a decline–regrowth hierarchical model predicted the next 12 weeks of PSA better than linear or single-decay models at every landmark (MAE 0.429 vs 0.543–0.565 at 6 weeks).
- Its advantage over simply carrying the last PSA value forward was uncertain at 6 and 12 weeks, and at 24 weeks the last value was more accurate (MAE 0.249 vs 0.327).
- The regrowth rate was weakly identified, with its population mean at the lower bound at every landmark, so no minimum observation period for a regrowth-based decision can be set from these data.
When patients stop being measured: how much does dropout change a tumor-shrinkage estimate?
Most tumor-scan dropout was death. A mixed model removed most of the survivor bias in 48-week shrinkage; a joint model and delta adjustment showed the 24-week arm difference was robust but the 48-week difference was fragile.
- In a public colorectal-cancer subset where most scan dropout was death, the mean of patients still scanned overstated 48-week shrinkage by about a third in the combination arm; a mixed-effects model removed most of that bias.
- The 24-week shrinkage and arm difference held under every dropout assumption tried, including a joint model and delta adjustments.
- The 48-week arm difference lost support if dropouts in the better-looking arm deteriorated about 0.5 per year faster than predicted; without recorded dropout reasons, the mechanism is not identified.
Do early tumor measurements improve prediction of subsequent survival?
A patient-separated landmark analysis of public colorectal cancer data, including calibration, censoring, uncertainty and temporal-integrity sensitivity.
- In a public 150-patient colorectal-cancer subset, a week-16 landmark analysis (128 patients) used only tumor measurements available before the prediction time.
- Adding early tumor change raised held-out concordance from 0.532 to 0.581 (difference +0.049, 95% interval −0.005 to +0.106), and calibration stayed weak (slope 0.464).
- Early tumor change is a candidate for further study, not a survival surrogate or decision rule; a separate validation cohort and denser longitudinal data are needed.
Does a tumor-kinetic model add anything to a single early tumor change?
At 16, 24 and 32 weeks, a mixed-effects tumor-kinetic estimate added no held-out survival discrimination beyond the observed one-point change, and carrying the last value forward predicted the next measurement best.
- At 16, 24 and 32 weeks, a mixed-effects tumor-kinetic estimate changed held-out survival C-index by only +0.002 to +0.016 over the observed one-point change, and every interval included zero.
- For the next tumor measurement, carrying the last value forward was the most accurate rule at every landmark (MAE 0.42–0.61), beating both linear and quadratic mixed models.
- Kinetic terms correlated about 0.97 with the one-point change, so with this measurement density a simple change suffices; this is not an evaluation of mechanistic TGI models in general.
When does individual information improve PK/PD prediction?
A reanalysis of warfarin PK/PCA data, revised with an absorption lag and a Hill exponent: effect delay, held-out prediction and the limits of extra PK.
- In 32 subjects with public single-dose warfarin PK and PCA data, a turnover model described the delay between concentration and response better than direct inhibition (held-out RMSE 0.0827 vs 0.0987 at the 24 h cutoff); version 1.1 added an absorption lag and version 1.2 a Hill exponent (1.28).
- Adding each subject's PK measurements up to 24 or 48 h did not improve later PCA prediction: paired MSE differences were +0.0077 (0.0003 to 0.0174) and +0.0051 (−0.0013 to 0.0114), worse at 24 h.
- The Hill exponent improved the in-sample fit but not held-out prediction. All current refits start from generic values; nothing here supports a dose, INR target or bleeding-risk claim.