Deep learning has become a powerful tool in solving image classification problems¹ and has been widely used in medical image analysis.² Convolutional neural networks (CNNs) show the potential for automated classification of different cancer lesions, such as lung cancer,³ skin cancer,⁴ and breast cancer.⁵ But the lack of interpretability and explainability makes CNNs less understandable.⁶ To push the envelope of CNN-based medical diagnostic methods, more rigorous neural network models and training methods must be developed to provide both visually interpretable and explainable analysis.^{7 – 9}

Secondary prevention in coronary heart disease is the prevention of occurrence of recurrent coronary events after clinical diagnosis. High level of adherence to secondary prevention interventions, especially aggressive lifestyle changes and pharmacotherapy can lead to significant decline in recurrent coronary events. Both international and Indian studies have reported low adherence to such therapies. Evidence-based useful interventions include regular physical activity, yoga, intake of healthy diet, smoking and tobacco use cessation and weight management. Pharmacotherapeutic interventions useful are anti-platelet therapy, target oriented lipid lowering therapy with statins, beta blockers and angiotensin converting enzyme inhibitors in patients with impaired left ventricular function. Hypertension and diabetes management with control to targets is important. Novel strategies include use of anticoagulants, anti-inflammatory drugs, and triglyceride lowering for residual risk. Physician and patient level interventions using multifaceted educational, socioeconomic and technological innovations are important to promote life-long adherence to these strategies.

Oral cancer is a common disease, particularly in low- and middle-income countries. Early detection of oral cancer is believed to be the most effective way to prevent it. Automatic oral cancer image classification algorithms based on machine learning enable a system to learn from previous data and, based on the learning, predict and give results on new unseen data. However, oral cancer datasets captured from high-risk populations are often unbalanced since there are many more normal cases than benign, premalignant, and malignant cases. Due to the influence of the larger majority classes featured on machine learning training framework, the classifier results may be biased toward the over-represented class (or classes). The discrimination of minority classes is also of great clinical significance. For example, a false positive result for benign lesions in cancer screening will result in unnecessary psychological stress, medical procedures to patients, and increased clinical workloads. Similarly, classifiers also need high sensitivity because they are designed for cancer screening. Therefore, one of the big challenges is handling class imbalance, especially for multiple category classification.

Diabetes and COVID-19: Role of insulin resistance as a risk factor for COVID-19 severity

Background: In the POISE-2 (PeriOperative ISchemic Evaluation 2) trial, perioperative aspirin did not reduce cardiovascular events, but increased major bleeding. There remains uncertainty regarding the effect of perioperative aspirin in patients undergoing vascular surgery. The aim of this substudy was to determine whether there is a subgroup effect of initiating or continuing aspirin in patients undergoing vascular surgery. Methods: POISE-2 was a blinded, randomized trial of patients having non-cardiac surgery. Patients were assigned to perioperative aspirin or placebo. The primary outcome was a composite of death or myocardial infarction at 30 days. Secondary outcomes included: vascular occlusive complications (a composite of amputation and peripheral arterial thrombosis) and major or life-threatening bleeding. Results: Of 10 010 patients in POISE-2, 603 underwent vascular surgery, 319 in the continuation and 284 in the initiation stratum. Some 272 patients had vascular surgery for occlusive disease and 265 had aneurysm surgery. The primary outcome occurred in 13·7 per cent of patients having aneurysm repair allocated to aspirin and 9·0 per cent who had placebo (hazard ratio (HR) 1·48, 95 per cent c.i. 0·71 to 3·09). Among patients who had surgery for occlusive vascular disease, 15·8 per cent allocated to aspirin and 13·6 per cent on placebo had the primary outcome (HR 1·16, 0·62 to 2·17). There was no interaction with the primary outcome for type of surgery (P = 0·294) or aspirin stratum (P = 0·623). There was no interaction for vascular occlusive complications (P = 0·413) or bleeding (P = 0·900) for vascular compared with non-vascular surgery. Conclusion: This study suggests that the overall POISE-2 results apply to vascular surgery. Perioperative withdrawal of chronic aspirin therapy did not increase cardiovascular or vascular occlusive complications. Registration number: NCT01082874 (http://www.clinicaltrials.gov).

Background: In previous analyses, myocardial injury after noncardiac surgery, major bleeding, and sepsis were independently associated with most deaths in the 30 days after noncardiac surgery, but most of these deaths occurred during the index hospitalization for surgery. The authors set out to describe outcomes after discharge from hospital up to 1 yr after inpatient noncardiac surgery and associations between predischarge complications and postdischarge death up to 1 yr after surgery. Methods: This study was an analysis of patients discharged after inpatient noncardiac surgery in a large international prospective cohort study across 28 centers from 2007 to 2013 of patients aged 45 yr or older followed to 1 yr after surgery. The study estimated (1) the cumulative postdischarge incidence of death and other outcomes up to a year after surgery and (2) the adjusted time-varying associations between postdischarge death and predischarge complications including myocardial injury after noncardiac surgery, major bleeding, sepsis, infection without sepsis, stroke, congestive heart failure, clinically important atrial fibrillation or flutter, amputation, venous thromboembolism, and acute kidney injury managed with dialysis. Results: Among 38,898 patients discharged after surgery, the cumulative 1-yr incidence was 5.8% (95% CI, 5.5 to 6.0%) for all-cause death and 24.7% (95% CI, 24.2 to 25.1%) for all-cause hospital readmission. Predischarge complications were associated with 33.7% (95% CI, 27.2 to 40.2%) of deaths up to 30 days after discharge and 15.0% (95% CI, 12.0 to 17.9%) up to 1 yr. Most of the association with death was due to myocardial injury after noncardiac surgery (15.6% [95% CI, 9.3 to 21.9%] of deaths within 30 days, 6.4% [95% CI, 4.1 to 8.7%] within 1 yr), major bleeding (15.0% [95% CI, 8.3 to 21.7%] within 30 days, 4.7% [95% CI, 2.2 to 7.2%] within 1 yr), and sepsis (5.4% [95% CI, 2.2 to 8.6%] within 30 days, 2.1% [95% CI, 1.0 to 3.1%] within 1 yr). Conclusions: One in 18 patients 45 yr old or older discharged after inpatient noncardiac surgery died within 1 yr, and one quarter were readmitted to the hospital. The risk of death associated with predischarge perioperative complications persists for weeks to months after discharge. Editor’s Perspective: What We Already Know about This Topic: What This Article Tells Us That Is New: