Comparative Efficacy of Drug Interventions for Keloids: A Network Meta-analysis.

BACKGROUND: Keloids are common benign skin lesions originating from a disorganized fibroproliferative collagen response; these lesions often lead to both physical and psychological problems. The optimal treatment for keloids is yet to be standardized. Intralesional injection, which is simple and nontraumatic, is one of the most commonly used treatment modalities for these lesions. In this study, we compared 5 different drugs (intralesional injections) for the treatment of keloids in terms of efficacy. METHODS: We systemically searched relevant studies on PubMed, EMBASE, and Cochrane Library. Randomized clinical trials on the safety and efficacy of triamcinolone acetonide (TAC), 5-fluorouracil (5-FU), botulinum toxin A (BTA), verapamil, and bleomycin were included in this study. RESULTS: This network meta-analysis included a total of 1114 patients from 20 randomized controlled trials. Botulinum toxin A alone and TAC plus 5-FU exhibited significantly better efficacy than did 5-FU, TAC, and verapamil. No significant difference in efficacy between BTA alone and TAC combined with 5-FU was observed. No significant differences were noted in the adverse event rate between BTA, TAC plus 5-FU, 5-FU, and TAC. Furthermore, we performed surface under the cumulative ranking curve analyses to predict the rank of each intervention (by efficacy and adverse event rate). The predicted ranking by efficacy was as follows: TAC plus 5-FU, BTA, bleomycin, TAC, 5-FU, and verapamil; the predicted ranking by adverse events was as follows: TAC, 5-FU, TAC plus 5-FU, and BTA. Funnel plot analysis revealed no publication bias. CONCLUSIONS: Botulinum toxin A and TAC plus 5-FU appear to have outstanding therapeutic efficacy for keloids. The rate of adverse events was similar among BTA, TAC, 5-FU, and TAC plus 5-FU. Nonetheless, additional reviews of rigorous, large-scale randomized controlled trials are warranted for further validation of our findings.

Direct prediction of carbapenem-resistant, carbapenemase-producing, and colistin-resistant Klebsiella pneumoniae isolates from routine MALDI-TOF mass spectra using machine learning and outcome evaluation.

The objective of this study was to develop a rapid prediction method for carbapenem-resistant Klebsiella pneumoniae (CRKP) and colistin-resistant K. pneumoniae (ColRKP) based on routine MALDI-TOF mass spectrometry (MS) results in order to formulate a suitable and rapid treatment strategy. A total of 830 CRKP and 1462 carbapenem-susceptible K. pneumoniae (CSKP) isolates were collected; 54 ColRKP isolates and 1592 colistin-intermediate K. pneumoniae (ColIKP) isolates were also included. Routine MALDI-TOF MS, antimicrobial susceptibility testing, NG-Test CARBA 5, and resistance gene detection were followed by machine learning (ML). Using the ML model, the accuracy and area under the curve for differentiating CRKP and CSKP were 0.8869 and 0.9551, respectively, and those for ColRKP and ColIKP were 0.8361 and 0.8447, respectively. The most important MS features of CRKP and ColRKP were m/z 4520-4529 and m/z 4170-4179, respectively. Of the CRKP isolates, MS m/z 4520-4529 was a potential biomarker for distinguishing KPC from OXA, NDM, IMP, and VIM. Of the 34 patients who received preliminary CRKP ML prediction results (by texting), 24 (70.6%) were confirmed to have CRKP infection. The mortality rate was lower in patients who received antibiotic regimen adjustment based on the preliminary ML prediction (4/14, 28.6%). In conclusion, the proposed model can provide rapid results for differentiating CRKP and CSKP, as well as ColRKP and ColIKP. The combination of ML-based CRKP with preliminary reporting of results can help physicians alter the regimen approximately 24 h earlier, resulting in improved survival of patients with timely antibiotic intervention.

Predicting microvascular invasion in hepatocellular carcinoma: a deep learning model validated across hospitals.

BACKGROUND: The accuracy of estimating microvascular invasion (MVI) preoperatively in hepatocellular carcinoma (HCC) by clinical observers is low. Most recent studies constructed MVI predictive models utilizing radiological and/or radiomics features extracted from computed tomography (CT) images. These methods, however, rely heavily on human experiences and require manual tumor contouring. We developed a deep learning-based framework for preoperative MVI prediction by using CT images of arterial phase (AP) with simple tumor labeling and without the need of manual feature extraction. The model was further validated on CT images that were originally scanned at multiple different hospitals. METHODS: CT images of AP were acquired for 309 patients from China Medical University Hospital (CMUH). Images of 164 patients, who took their CT scanning at 54 different hospitals but were referred to CMUH, were also collected. Deep learning (ResNet-18) and machine learning (support vector machine) models were constructed with AP images and/or patients' clinical factors (CFs), and their performance was compared systematically. All models were independently evaluated on two patient cohorts: validation set (within CMUH) and external set (other hospitals). Subsequently, explainability of the best model was visualized using gradient-weighted class activation map (Grad-CAM). RESULTS: The ResNet-18 model built with AP images and patients' clinical factors was superior than other models achieving a highest AUC of 0.845. When evaluating on the external set, the model produced an AUC of 0.777, approaching its performance on the validation set. Model interpretation with Grad-CAM revealed that MVI relevant imaging features on CT images were captured and learned by the ResNet-18 model. CONCLUSIONS: This framework provide evidence showing the generalizability and robustness of ResNet-18 in predicting MVI using CT images of AP scanned at multiple different hospitals. Attention heatmaps obtained from model explainability further confirmed that ResNet-18 focused on imaging features on CT overlapping with the conditions used by radiologists to estimate MVI clinically.