Machine learning to predict high-dose methotrexate-related neutropenia and fever in children with B-cell acute lymphoblastic leukemia

Methotrexate (MTX), an antimetabolite for the treatment of leukemia, could cause neutropenia and subsequently fever, which might lead to treatment delay and affect prognosis. Here, we aimed to predict neutropenia and fever related to high-dose MTX using artificial intelligence. This study included 139 pediatric patients newly diagnosed with standard- or intermediate risk B-cell acute lymphoblastic leukemia. Fifty-seven SNPs of 16 genes were genotyped. Univariate and multivariate analysis were used to select SNPs and clinical covariates for model developing. Five machine learning algorithms combined with four resampling techniques were used to build optimal predictive model. The combination of random forest with adaptive synthetic appeared to be the best model for neutropenia (sensitivity = 0.935, specificity = 0.920, AUC = 0.927) and performed best for fever (sensitivity = 0.818, specificity = 0.924, AUC = 0.870). By machine learning, we have developed and validated comprehensive models to predict the risk of neutropenia and fever. Such models may be helpful for medical oncologists in quick decision-making.

[Predictive value of four score systems on the bleeding risk in atrial fibrillation patients treated with dabigatran]

Objective: To compare the predictive value of HAS-BLED, HEMORR₂HAGES, ATRIA and ORBIT scores on the bleeding risk in nonvalvular atrial fibrillation (NVAF) patients treated with dabigatran. Methods: Data of 942 NVAF patients participating a non-interventional prospective study of anticoagulant therapy with dabigatran, which was conducted in 12 centers from February 2015 to December 2017 in China, were analyzed. Complete HAS-BLED HEMORR₂HAGES, ATRIA and ORBIT bleeding risk scores data and follow-up data were available in the enrolled patients. The endpoint of the study was bleeding events occurred during a 6 months follow-up. Cox proportional hazards models were constructed to analyze the associations between HAS-BLED, HEMORR₂HAGES, ATRIA and ORBIT scores and risk of bleeding, and the area under the curve (AUC) of receiver operating characteristics curves (ROC) of each score was used to set the predictive value for bleeding risk. Results: Among the 942 patients, the mean age was (65.3±11.2) years old, 542 (57.5%) were males. A total of 93 (9.9%) bleeding events occurred during follow up, 89 (9.4%) events were minor bleeding, and 4 (0.4%) events were major bleeding. Patients with a high-risk HAS-BLED score had a 1.87-fold increased risk of bleeding compared with low-risk patients (HR = 2.87, 95% CI:1.26-6.51, P = 0.012). There was no statistically significant difference between low-medium-high-risk grading in other scoring systems and bleeding risk (all P>0.05). The AUC (95%CI) of HAS-BLED, HEMORR₂HAGES, ATRIA and ORBIT bleeding risk scores were 0.558 (0.525-0.590), 0.520 (0.487-0.553), 0.513(0.480-0.545), 0.523(0.490-0.555), respectively. The AUC of all bleeding score systems were of ≤ 0.700. Conclusion: Among the NVAF patients taking dabigatran in China, the HAS-BLED bleeding risk score is superior to other 3 bleeding risk score on predicting the bleeding risk in these patients, but its predictive value is still relatively low.

Analysis on the local structures for 3d1 impurities (Ti3+ and V4+ ) in KTiPO4

Making use of the perturbation formulae for 3d¹ ions (Ti³⁺ and V⁴⁺ ) under orthorhombically compressed octahedra, the spin Hamiltonian parameters (g factors: g_x , g_y , g_z and hyperfine structure constants: A_x , A_y , A_z ) and local structures of the 3d¹ impurity centres C₁ , C₂ , and C₃ in KTiOPO₄ crystals are theoretically analyzed in a consistent way. The remarkable local distortions (i.e., the relative axial compression ratios 11.2%, 7.0%, and 5.5% along Z axis and the relative planar bond length variation ratios 15.9%, 7.0%, and 6.0%) are obtained for the [Ti2O₆ ]^9- cluster on Ti2 site and [VO₆ ]^8- clusters on Ti1 and Ti2 sites, respectively, in view of the Jahn-Teller effect. The above local orthorhombic distortion parameters in the impurity centres are found to be more significant than the host Ti1 and Ti2 sites in pure KTiOPO₄ . The sequences (C₁  > C₂  > C₃ ) of the local orthorhombic distortion parameters ρ and τ are in accordance with those of the axial and perpendicular anisotropies Δg and δg of g factors, respectively.