Stratified Intervention Assignment

See Stratified Random Sampling Task, Stratified K-Fold Cross-Validation, Stratified Psychiatry.

References

(Joyce et al., 2017) ⇒ Dan W. Joyce, Angie A. Kehagia, Derek K. Tracy, Jessica Proctor, and Sukhwinder S. Shergill (2017). "Realising stratified psychiatry using multidimensional signatures and trajectories". Journal of translational medicine, 15(1), 15. DOI:10.1186/s12967-016-1116-1
- QUOTE: For example, given a large number of patients whose PANSS scores (a 30-item instrument) are used to populate a 30-dimensional space, we might seek a set of prototypes that represent clusters of patients that are sufficiently similar such that stratified assignment to intervention occurs by similarity in psychopathology. Of note, this represents exploration and discovery of candidate prototypes and the model is constructed by ‘unsupervised’ learning. In Fig. 1b, stratification of a new, previously unknown patient would be some function of the distance to discovered prototypes (solid green and red lines)

**Figure 1:** a Schematic representation of two patients’ signatures at time 1 (before intervention) and time 2 (after intervention) with connecting lines representing trajectory. b A space of patient signatures along dimensions of positive and negative symptoms, with black and grey points representing patients belonging to two tentative clusters of patients. Diamonds show the centroids (prototypes) of the two patient categories. Dotted lines represent between patient-signature similarity and solid lines represent class membership as proportional to the distance of a given patient to the nearest prototype/centroid. The blue solid line represents a linear discriminant separating the two classes such that the overall misclassification error is minimised, as might be estimated or learned by, for example, a support vector machine or linear discriminant analysis.