OBJECTIVE To provide guidance on the scientific and standardized application of artificial intelligence （AI） and multimodal fusion techniques in constructing predictive models for adverse drug reactions （ADR） of antineoplastics， thereby enhancing the safety of antineoplastic therapy. METHODS Jointly initiated by the Clinical Pharmacy and Translational Pharmacy Society of the Liaoning Provincial Life Sciences Association and other professional societies， and organized by the Fourth Affiliated Hospital of China Medical University， the writing group conducted a comprehensive search of domestic and international databases using keywords such as “multimodal data” “artificial intelligence” “adverse drug reactions”“antineoplastics” and “models” to collect guidelines， expert consensuses， and relevant guidance literature. Based on this search， and combined with China’s practical experience in applying AI to oncology pharmaceutical care， a consensus framework and preliminary recommendations were drafted. Following evaluation and discussion by an external expert panel， the writing group incorporated the assessment results and formulated recommendations based on those endorsed by 90% or more of the experts. RESULTS &CONCLUSIONS After multiple rounds of review， discussion and revision， this consensus outlines the process for constructing an ADR prediction model of antineoplastics， encompassing “defining research objectives-data preparation-selection of algorithmic frameworks-design of multimodal data fusion strategies-model training and validation-interpretability analysis-clinical deployment-publication of results”. It also formulates 18 recommendations regarding core aspects such as data preparation， algorithm framwork selection， multimodal data fusion strategy design， model validation and interpretability analysis， and clinical deployment. These recommendations are intended for use by physicians in the field of oncology， clinical pharmacists， researchers， and AI technology developers to assist in making optimal clinical treatment decisions.

BACKGROUND: This study aimed to develop a comprehensive instrument for evaluating and ranking clinical practice guidelines, named Scientific, Transparent and Applicable Rankings tool (STAR), and test its reliability, validity, and usability. METHODS: This study set up a multidisciplinary working group including guideline methodologists, statisticians, journal editors, clinicians, and other experts. Scoping review, Delphi methods, and hierarchical analysis were used to develop the STAR tool. We evaluated the instrument's intrinsic and interrater reliability, content and criterion validity, and usability. RESULTS: STAR contained 39 items grouped into 11 domains. The mean intrinsic reliability of the domains, indicated by Cronbach's α coefficient, was 0.588 (95% confidence interval [CI]: 0.414, 0.762). Interrater reliability as assessed with Cohen's kappa coefficient was 0.774 (95% CI: 0.740, 0.807) for methodological evaluators and 0.618 (95% CI: 0.587, 0.648) for clinical evaluators. The overall content validity index was 0.905. Pearson's r correlation for criterion validity was 0.885 (95% CI: 0.804, 0.932). The mean usability score of the items was 4.6 and the median time spent to evaluate each guideline was 20 min. CONCLUSION The instrument performed well in terms of reliability, validity, and efficiency, and can be used for comprehensively evaluating and ranking guidelines.
Reproducibility of Results ; Surveys and Questionnaires ; Practice Guidelines as Topic ; Humans

Transparency Ecosystem for Research and Journals in Medicine (TERM) working group summarized the essential recommendations that should be considered to review and publish a high-quality guideline. These recommendations from editors and reviewers included 10 components of essential requirements: systematic review of existing relevant guidelines, guideline registration, guideline protocol, stakeholders, conflicts of interest, clinical questions, systematic reviews, recommendation consensus, guideline reporting and external review. TERM working group abbreviates them as PAGE (essential requirements for Publishing clinical prActice GuidelinEs), and recommends guideline authors, editors, and peer reviewers to use them for high-quality guidelines.
Humans ; Practice Guidelines as Topic

Chromosome Aberrations ; Consensus ; Cytogenetics ; Humans ; Multiple Myeloma

Four trials of external quality assessment in diagnostic hematology were performed in 2009 with average 946 participating laboratories in Korea. We performed quality assessment for white blood cell count, hemoglobin, hematocrit, red blood cell count, platelet count, blood cell morphology, prothrombin time and activated partial thromboplastin time. The response rate was more than 98.2%. The coefficients of variation in hemoglobin, hematocrit and RBC count were stable but variable in platelet count and WBC count according to measuring cell counters. Test results of blood cell morphology showed variation among various cell morphologies.
Blood Cells ; Cell Count ; Erythrocyte Count ; Hematocrit ; Hematology ; Hemoglobins ; Korea ; Leukocyte Count ; Partial Thromboplastin Time ; Platelet Count ; Prothrombin Time

Four trials of external quality assessment in diagnostic hematology were performed in 2008 with average 822 participating laboratories in Korea. We performed quality assessment for white blood cell count, hemoglobin, hematocrit, red blood cell count, platelet count, blood cell morphology, prothrombin time and activated partial thromboplastin time. The response rate was more than 96.5%. The coefficients of variation in hemoglobin, hematocrit and RBC was stable but variable in platelet count and WBC count according to measuring cell count. Test results of blood cell morphology showed variation among various cell morphologies.
Blood Cells ; Cell Count ; Erythrocyte Count ; Hematocrit ; Hematology ; Hemoglobins ; Korea ; Leukocyte Count ; Partial Thromboplastin Time ; Platelet Count ; Prothrombin Time

Four trials of external quality assessment in diagnostic hematology were performed in 2007 with average 722 participating laboratories in Korea. We performed quality assessment for white blood cell count, hemoglobin, hematocrit, red blood cell count, platelet count, blood cell morphology, prothrombin time and activated partial thromboplastin time. The response rate was more than 95.2%. The coefficients of variation in hemoglobin, hematocrit and RBC were stable but variable in platelet count and WBC count according to measuring cell counters. Test results of blood cell morphology showed variation among various cell morphologies.
Blood Cells ; Cell Count ; Erythrocyte Count ; Hematocrit ; Hematology ; Hemoglobins ; Korea ; Leukocyte Count ; Partial Thromboplastin Time ; Platelet Count ; Prothrombin Time