Table 7.

Study analysis for journal publications on the treatment phase.

Reference	Objective, data set, and methodology	Performance and remarks
[9]	Objective: Digital analysis of blood smears and preclassification of cells Data set: Images of blood smears from a hematologic laboratory Methodology: MERGE algorithm	Performance: Accuracy: 90% Strengths: Introduction of a new computational and statistical method to determine gene markers Limitations: Small data set comprising only 30 patients with acute myeloid leukemia Validation: Leave-one-out cross validation
[75]	Objective: Prediction of complete remission of acute myeloid leukemia Data set: 473 bone marrow samples from the Children’s Oncology Group Methodology: K-nearest neighbor, support vector machine, and hill climbing	Performance: Area under the curve: 0.84 Strengths: Use of 3 feature selection algorithms: randomized LASSO, recursive feature elimination, and hill climbing Use of 3 classifiers: support vector machine, random forest, and K-nearest neighbor Limitations: Small data set Validation: 100 iterations of a 5-fold cross validation
[81]	Objective: Identify the right patterns to improve risk stratification of patient with CLLsa Data set: (1) the first cohort comprised CLL cells of 196 individuals; the second cohort comprised CLL cells of 98 individuals including their clinical data and RNA-seq Methodology: (1) EM algorithm and the Gaussian mixture models; (2) Boosted tree ensemble method	Performance: Precision: 90% Strengths: High accuracy and precision Limitations: Large data set and 5-year monitoring is required Validation: External validation on an independent cohort

^aCLL: chronic lymphocytic leukemia.