Risk Prediction of Diabetic Foot Amputation Using Machine Learning and Explainable Artificial Intelligence.

BACKGROUND: Diabetic foot ulcers (DFUs) are serious complications of diabetes which can lead to lower extremity amputations (LEAs). Risk prediction models can identify high-risk patients who can benefit from early intervention. Machine learning (ML) methods have shown promising utility in medical applications. Explainable modeling can help its integration and acceptance. This study aims to develop a risk prediction model using ML algorithms with explainability for LEA in DFU patients. METHODS: This study is a retrospective review of 2559 inpatient DFU episodes in a tertiary institution from 2012 to 2017. Fifty-one features including patient demographics, comorbidities, medication, wound characteristics, and laboratory results were reviewed. Outcome measures were the risk of major LEA, minor LEA and any LEA. Machine learning models were developed for each outcome, with model performance evaluated using receiver operating characteristic (ROC) curves, balanced-accuracy and F1-score. SHapley Additive exPlanations (SHAP) was applied to interpret the model for explainability. RESULTS: Model performance for prediction of major, minor, and any LEA event achieved ROC of 0.820, 0.637, and 0.756, respectively, with XGBoost, XGBoost, and Gradient Boosted Trees algorithms demonstrating best results for each model, respectively. Using SHAP, key features that contributed to the predictions were identified for explainability. Total white cell (TWC) count, comorbidity score and red blood cell count contributed highest weightage to major LEA event. Total white cell, eosinophils, and necrotic eschar in the wound contributed most to any LEA event. CONCLUSIONS: Machine learning algorithms performed well in predicting the risk of LEA in a patient with DFU. Explainability can help provide clinical insights and identify at-risk patients for early intervention.

Development of an explainable artificial intelligence model for Asian vascular wound images.

Chronic wounds contribute to significant healthcare and economic burden worldwide. Wound assessment remains challenging given its complex and dynamic nature. The use of artificial intelligence (AI) and machine learning methods in wound analysis is promising. Explainable modelling can help its integration and acceptance in healthcare systems. We aim to develop an explainable AI model for analysing vascular wound images among an Asian population. Two thousand nine hundred and fifty-seven wound images from a vascular wound image registry from a tertiary institution in Singapore were utilized. The dataset was split into training, validation and test sets. Wound images were classified into four types (neuroischaemic ulcer [NIU], surgical site infections [SSI], venous leg ulcers [VLU], pressure ulcer [PU]), measured with automatic estimation of width, length and depth and segmented into 18 wound and peri-wound features. Data pre-processing was performed using oversampling and augmentation techniques. Convolutional and deep learning models were utilized for model development. The model was evaluated with accuracy, F1 score and receiver operating characteristic (ROC) curves. Explainability methods were used to interpret AI decision reasoning. A web browser application was developed to demonstrate results of the wound AI model with explainability. After development, the model was tested on additional 15 476 unlabelled images to evaluate effectiveness. After the development on the training and validation dataset, the model performance on unseen labelled images in the test set achieved an AUROC of 0.99 for wound classification with mean accuracy of 95.9%. For wound measurements, the model achieved AUROC of 0.97 with mean accuracy of 85.0% for depth classification, and AUROC of 0.92 with mean accuracy of 87.1% for width and length determination. For wound segmentation, an AUROC of 0.95 and mean accuracy of 87.8% was achieved. Testing on unlabelled images, the model confidence score for wound classification was 82.8% with an explainability score of 60.6%. Confidence score was 87.6% for depth classification with 68.0% explainability score, while width and length measurement obtained 93.0% accuracy score with 76.6% explainability. Confidence score for wound segmentation was 83.9%, while explainability was 72.1%. Using explainable AI models, we have developed an algorithm and application for analysis of vascular wound images from an Asian population with accuracy and explainability. With further development, it can be utilized as a clinical decision support system and integrated into existing healthcare electronic systems.

Use of Wallstent device as an embolic protection device during stenting of aortic thrombus.

Blue toe syndrome can occur due to distal embolization from proximal lesions such as an aortic thrombus. We describe the case of a patient who presented with chronic limb threatening ischemia due to a flow-limiting infrarenal aortic thrombus, with gangrene from distal embolization to the left fifth toe, and was successfully treated with endovascular aortic stent graft insertion. Distal embolization during instrumentation was successfully prevented by using a partially deployed Wallstent (Boston Scientific) as an embolic protection device. The reconstrainable Wallstent device can be considered for distal thromboembolic protection during aortic stenting, in particular, when distal embolization is a concern and commercial devices are not readily available.

Pancreaticoduodenectomy after Roux-en-Y Gastric Bypass: a novel reconstruction technique.

The obesity epidemic continues to increase around the world with its attendant complications of metabolic syndrome and increased risk of malignancies, including pancreatic malignancy. The Roux-en-Y gastric bypass (RYGB) is an effective bariatric procedure for obesity and its comorbidities. We describe a report wherein a patient with previous RYGB was treated with a novel reconstruction technique following a pancreaticoduodenectomy (PD). A 59-year-old male patient with previous history of RYGB was admitted with painless progressive jaundice. Imaging revealed a distal common bile duct stricture and he underwent PD. There are multiple options for reconstruction after PD in patients with previous RYGB. The two major decisions for pancreatic surgeon are: (I) resection/preservation of remnant stomach and (II) resection/preservation of original biliopancreatic limb. This has to be tailored to the patient based on the intraoperative findings and anatomical suitability. In our patient, the gastric remnant was preserved, and distal part of original biliopancreatic limb was anastomosed to the stomach as a venting anterior gastrojejunostomy. A distal loop of small bowel was used to reconstruct the pancreaticojejunostomy and hepaticojejunostomy and further distally a new jejunojejunostomy performed. The post-operative course was uneventful, and the patient was discharged on 7th day. With the increase in number of bariatric procedures performed worldwide, pancreatic surgeons should be aware of the varied surgical reconstruction options for PD following RYGB. This should be tailored to the patient and there is no "one-size-fits-all".

A prospective validation of Sepsis-3 guidelines in acute hepatobiliary sepsis: qSOFA lacks sensitivity and SIRS criteria lacks specificity (Cohort Study).

BACKGROUND: Since its introduction in 2016, the Sepsis-3 guidelines, with emphasis on the quick Sequential Organ Failure Assessment (qSOFA) score, have generated much debate and controversy. It is recognised that the new definitions require validation in specific clinical settings and have yet to be universally adopted. We aim to validate new Sepsis-3 guidelines in acute hepatobiliary infection. MATERIAL AND METHODS: A prospective cohort of patients admitted with acute hepatobiliary infection from the emergency department from July 2016 to June 2017 was studied. The Systemic Inflammatory Response Syndrome (SIRS) criteria, SOFA and qSOFA scores were calculated and predictive performance evaluated with area under the receiver operating characteristic (AUROC) curves for predictive ability of these indices for critical care unit admission and morbidity. RESULTS: 124 patients with a median age of 64.5 years and majority males (n = 75, 60.5%) were admitted with acute hepatobiliary infection during the study period. Acute cholecystitis was the most common admission diagnosis (n = 83, 66.9%) and most patients were managed in general ward (n = 91, 73.3%) with median length of stay of 6 days (range 1-40). On multivariate analysis, diabetes mellitus (p = 0.003) predicted high dependency unit (HDU) admission, while age (p = 0.001), positive blood culture (p = 0.012), positive fluid culture (p = 0.015) and SOFA score (p = 0.002) predicted length of hospital stay. The sensitivity of SIRS in predicting HDU admission (60% vs. 4%), intensive care unit (ICU) admission (62.5% vs. 0%) and morbidity (66.7% vs. 0%) was higher than qSOFA score. The specificity of qSOFA in predicting HDU admission (100% vs. 49.5%), ICU admission (99.1% vs. 53.3%) and morbidity (99.2% vs. 47.9%) was higher than SIRS criteria. CONCLUSION: The SIRS criteria has high sensitivity and the qSOFA score has high specificity in predicting outcomes of patients with acute hepatobiliary infection.

Congenital or acquired? Obstructive jaundice in reoperated duodenal atresia.

A 55-day-old boy was transferred to our unit with intestinal obstruction and obstructive jaundice after two neonatal operations for duodenal atresia and intestinal malrotation. Abdominal ultrasound showed dilated intrahepatic and extrahepatic ducts with cut-off at the distal common bile duct (CBD). He underwent emergency laparotomy for adhesive intestinal obstruction with a contained abscess from mid-jejunal perforation. Biliary dissection was not attempted due to poor preoperative nutritional status. Tube cholecystostomy was created for biliary decompression. Postoperative magnetic resonance cholangiopancreatography showed dilated CBD with cut-off at the ampulla but did not demonstrate pancreaticobiliary maljunction (PBMJ). The diagnostic dilemma was whether our patient had congenital PBMJ or had developed biliary stricture from perioperative ischaemic scarring. He underwent definitive surgery at 7 months: excision of dilated CBD with Roux-en-Y hepaticojejeunal reconstruction, excisional tapering duodenoplasty and jejunostomy creation. Intraoperative finding was type I choledochal cyst and subsequently confirmed on histology. Postoperative recovery was uneventful and bilirubin levels normalised.