Objective To propose strategies for developing clinical predictive models,aiming to assist researchers in conducting standardized clinical prediction model studies.Methods Literature review was conducted to summarize the operational steps and content for developing clinical predictive models.Then,a methodological framework was summarized and refined through expert consultation.Results The 11-step methodological framework for developing clinical predictive models was obtained by synthesizing the experience of 456 clinical predictive modeling studies and expert consultation,and the details were analyzed and elaborated.Conclusions This study presents methodological strategies and recommendations for the development of clinical predictive models,intended to serve as a guide for researchers.

OBJECTIVE： To study the relationships of polymorphism of MTRR gene rs1801394 locus and SLCO1B1 gene rs11045879 locus with drug concentration of methotrexate （MTX） and high-dose MTX （HD-MTX）-induced ADR in acute lymphoblastic leukemia （ALL） children. METHODS： From Oct. 2015 to Sept. 2018， 70 ALL hospitalized children of Han nationality in Sichuan area who received HD-MTX treatment and were in consolidation chemotherapy were selected retrospectively from Sichuan People’s Hospital. The blood concentration of MTX at 48 and 72 hours after administration was measured by EMIT. The genetic typing of MTRR gene rs1801394 locus and SLCO1B1 gene rs11045879 locus were detected with real-time PCR. The relationships of the polymorphism of MTRR gene and SLCO1B1 gene with MTX blood concentration [dose-corrected concentration （c48 h/D，48 h）， the proportion of children with different concentration of MTX （≤0.1， ＞0.1 μmol/L）] and ADR （such as myelosuppression， liver function damage， gastrointestinal response， mucosal damage， rash， etc.） were analyzed. Binary Logistic regression analysis for the correlation of ADR with different influencing factors （gene polymor- phism， blood concentration of MTX， immunophenotyping， body mass index， etc.） was carried out by Wald method. RESULTS： Totally 31， 32， 7 children with MTRR gene AA， AG and GG genotype， while 23， 37， 10 children with SLCO1B1 gene TT， TC and CC genotype were detected. The distribution of each genotype in 70 children conformed to Hardy-Weinberg equilibrium （P＞0.05）. There was no significant difference in c48 h/D（48 h） of children and the proportion of children with different concentration of MTX （72 h） among difterent genotypes of MTRR and SLCO1B1 gene （P＞0.05）. There was statistical significance in the incidence of liver function injury in children with different genotypes of MTRR gene （P＜0.05）， and the AA genotype was significantly higher than the AG+GG genotype （P＜0.05）. There was no correlation of MTRR gene polymorphism with the incidence of other ADR， neither SLCO1B1 gene polymorphism with the incidence of ADR （P＞0.05）. The results of Binary Logistic regression analysis showed that liver function damage in ALL children was related to the gene polymorphism of MTRR； gastrointestinal reaction was related to whether the plasma concentration was more than 0.1 μmol/L at 72 h； mucosal damage was related to the immune type and BMI of children； the occurrence of skin allergy was correlated with body weight of children（P＜0.05）. CONCLUSIONS： Gene polymorphism of MTRR rs1801394 locus may associated with the occurrence of HD-HTX-induced liver function injury in ALL children， but its polymorphism and gene polymorphism of SLCO1B1 rs11045879 locus are not related to MTX blood concentration in ALL children.