Search

Chronic wounds affect 6.5 million Americans, are complex conditions to manage and cost $28–$32 billion annually. Although digital solutions exist for non-expert clinicians to accurately segment tissues, analyze affected tissues or efficiently document their wound assessment results, there exists a lack of decision support for non-expert clinicians who usually provide most wound assessments and care decisions at the point of care (POC). We designed a machine learning (ML) system that can accurately predict wound care decisions based on labeled wound image data. The care decisions we predict are based on guidelines for standard wound care and are labeled as: continue the treatment, request a change in treatment, or refer patient to a specialist. In this paper, we demonstrate how our final ML solution using XGboost (XGB) algorithm achieved on average an overall performance of F-1 = .782 using labels given by an expert and a novice decision maker. The key contribution of our research lies in the ability of the ML artifact to use only those wound features (predictors) that require less expertise for novice users when examining wounds to make standard of care decisions (predictions).

Lower extremity chronic wounds affect 4.5 million Americans annually. Due to inadequate access to wound experts in underserved areas, many patients receive non-uniform, non-standard wound care, resulting in increased costs and lower quality of life. We explored machine learning classifiers to generate actionable wound care decisions about four chronic wound types (diabetic foot, pressure, venous, and arterial ulcers). These decisions (target classes) were: (1) Continue current treatment, (2) Request non-urgent change in treatment from a wound specialist, (3) Refer patient to a wound specialist. We compare classification methods (single classifiers, bagged & boosted ensembles, and a deep learning network) to investigate (1) whether visual wound features are sufficient for generating a decision and (2) whether adding unstructured text from wound experts increases classifier accuracy. Using 205 wound images, the Gradient Boosted Machine (XGBoost) outperformed other methods when using both visual and textual wound features, achieving 81% accuracy. Using only visual features decreased the accuracy to 76%, achieved by a Support Vector Machine classifier. We conclude that machine learning classifiers can generate accurate wound care decisions on lower extremity chronic wounds, an important step toward objective, standardized wound care. Higher decision-making accuracy was achieved by leveraging clinical comments from wound experts.

BACKGROUND: Diabetes mellitus continues to grow in global prevalence and to consume an increasing amount of health care resources. One of the key areas of morbidity associated with diabetes is the diabetic foot. To improve the care of patients with diabetic foot and to provide an evidence-based multidisciplinary management approach, the Society for Vascular Surgery in collaboration with the American Podiatric Medical Association and the Society for Vascular Medicine developed this clinical practice guideline. METHODS: The committee made specific practice recommendations using the Grades of Recommendation Assessment, Development, and Evaluation system. This was based on five systematic reviews of the literature. Specific areas of focus included (1) prevention of diabetic foot ulceration, (2) off-loading, (3) diagnosis of osteomyelitis, (4) wound care, and (5) peripheral arterial disease. RESULTS: Although we identified only limited high-quality evidence for many of the critical questions, we used the best available evidence and considered the patients' values and preferences and the clinical context to develop these guidelines. We include preventive recommendations such as those for adequate glycemic control, periodic foot inspection, and patient and family education. We recommend using custom therapeutic footwear in high-risk diabetic patients, including those with significant neuropathy, foot deformities, or previous amputation. In patients with plantar diabetic foot ulcer (DFU), we recommend off-loading with a total contact cast or irremovable fixed ankle walking boot. In patients with a new DFU, we recommend probe to bone test and plain films to be followed by magnetic resonance imaging if a soft tissue abscess or osteomyelitis is suspected. We provide recommendations on comprehensive wound care and various débridement methods. For DFUs that fail to improve (>50% wound area reduction) after a minimum of 4 weeks of standard wound therapy, we recommend adjunctive wound therapy options. In patients with DFU who have peripheral arterial disease, we recommend revascularization by either surgical bypass or endovascular therapy. CONCLUSIONS: Whereas these guidelines have addressed five key areas in the care of DFUs, they do not cover all the aspects of this complex condition. Going forward as future evidence accumulates, we plan to update our recommendations accordingly.

Background Despite advances in the treatment of peripheral arterial disease, a significant number of patients ultimately require major amputations. Traditionally, postoperative management of a below-knee amputation involves soft compressive dressings to allow for complete stump healing before initial prosthesis fitting. This technique is associated with a prolonged period of limited mobility, placing patients at risk for deconditioning or fall with a risk of injury to the stump. In contrast, immediate postoperative prosthesis (IPOP) placement allows patients to begin ambulation and rehabilitation on postoperative day 1, which may be of significant physiologic and psychological benefit. The purpose of this study is to compare the outcomes of patients undergoing IPOP placement to those of a historical control group managed with traditional soft compressive dressing placement. Methods Medical records of all consecutive below-knee amputation patients who underwent IPOP (IPOP group; 37 patients, 2007–2010) and all patients who underwent traditional soft compressive dressing placement and were IPOP candidates (non-IPOP group; 35 patients, 2006–2007) were retrospectively reviewed. Patient comorbidities and preoperative ambulation status were compared between the IPOP and the non-IPOP groups. Primary outcomes evaluated included perioperative systemic complications, wound complications, need for surgical revision, and the time until placement of a definitive prosthesis. Data were analyzed using the chi-squared and Student’s t -test. Results Preoperative comorbidities and patient characteristics of the 2 groups were similar, although the IPOP group was younger (61.5 vs. 69.0 years; P = 0.01). Immediate perioperative systemic complication rates were not significantly different between the 2 groups (IPOP 29.7% vs. non-IPOP 31.4%; P = 0.876). Postoperative wound complication rates were as follows: wound infection (IPOP 18.9% vs. non-IPOP 25.0%; P = 0.555), wound dehiscence (IPOP 29.7% vs. non-IPOP 25.0%; P = 0.673), and skin breakdown separate from the incision (IPOP 18.9% vs. non-IPOP 3.6%; P = 0.062). Patients in the IPOP group trended towards fewer postoperative falls (IPOP 10.8% vs. non-IPOP 21.4%; P = 0.240). The need for revision was significantly greater in the non-IPOP group (IPOP 5.4% vs. non-IPOP 27.6%; P = 0.013). The time from surgery to placement of the preparatory prosthesis was 51 days in the IPOP group. Conclusions Patients undergoing IPOP have similar perioperative systemic and wound complication rates compared to those patients undergoing conventional below-knee amputation, but are less likely to require surgical revision. The use of IPOP allows for early ambulation and rehabilitation, which may be of psychological benefit and may decrease the sequelae of prolonged immobilization. IPOP application should be considered for all appropriate candidates requiring below-knee amputation.

Objective The Society for Vascular Surgery Wound, Ischemia, foot Infection (WIfI) system aims to stratify threatened limbs according to their anticipated natural history and estimate the likelihood of benefit from revascularization, but whether it accurately stratifies outcomes in limbs undergoing aggressive treatment for limb salvage is unknown. We investigated whether the WIfI stage correlated with the intensity of limb treatment required and patient-centered outcomes. Methods We stratified limbs from a prospectively maintained database of consecutive patients referred to a limb preservation center according to WIfI stage (October 2013-May 2015). Comorbidities, multimodal limb treatment, including foot operations and revascularization, and patient-centered outcomes (wound healing, limb salvage, amputation-free survival, maintenance of ambulatory and independent living status, and mortality) were compared among WIfI stages. Multivariate analysis was performed to identify predictors of wound healing and limb salvage. Results We identified 280 threatened limbs encompassing all WIfI stages in 257 consecutive patients: stage 1, 48 (17%); stage 2, 67 (24%); stage 3, 64 (23%); stage 4, 83 (30%); and stage 5 (unsalvageable), 18 (6%). Operative foot debridement, minor amputation, and use of revascularization increased with increasing WIfI stage ( P ≤ .04). Revascularization was performed in 106 limbs (39%), with equal use of open and endovascular procedures. Over a median follow-up of 209 days (interquartile range, 95, 340) days, 1-year Kaplan-Meier wound healing cumulative incidence was 71%, and the proportion with complete wound healing decreased with increasing WIfI stage. Major amputation was required in 26 stage 1 to 4 limbs (10%). Increasing WIfI stage was associated with decreased 1-year Kaplan-Meier limb salvage (stage 1: 96%, stage 2: 84%, stage 3: 90%, and stage 4: 78%; P = .003) and amputation-free survival ( P = .006). Stage 4 WIfI independently predicted amputation (hazard ratio, 12; 95% confidence interval, 1.6-94). Amputation rates in patients with severe Ischemia grade 3 were lower in those who underwent revascularization than in those who did not (14% vs 41%; P = .01) Ambulatory and independent living status at follow-up deteriorated significantly from baseline in stage 4 but not stage 1 to 3 patients. Mortality was not different between WIfI stages. Conclusions In patients treated aggressively for limb salvage, WIfI stage correlated with intensity of multimodal limb treatment and with limb salvage and patient-centered outcomes at 1 year. Revascularization improved limb salvage in severe ischemia. These data support the Society for Vascular Surgery WIfI system as a powerful tool to risk-stratify patients with threatened limbs and guide treatment.