ROAD2H: Development and evaluation of an open-source explainable artificial intelligence approach for managing co-morbidity and clinical guidelines.

Introduction: Clinical decision support (CDS) systems (CDSSs) that integrate clinical guidelines need to reflect real-world co-morbidity. In patient-specific clinical contexts, transparent recommendations that allow for contraindications and other conflicts arising from co-morbidity are a requirement. In this work, we develop and evaluate a non-proprietary, standards-based approach to the deployment of computable guidelines with explainable argumentation, integrated with a commercial electronic health record (EHR) system in Serbia, a middle-income country in West Balkans. Methods: We used an ontological framework, the Transition-based Medical Recommendation (TMR) model, to represent, and reason about, guideline concepts, and chose the 2017 International global initiative for chronic obstructive lung disease (GOLD) guideline and a Serbian hospital as the deployment and evaluation site, respectively. To mitigate potential guideline conflicts, we used a TMR-based implementation of the Assumptions-Based Argumentation framework extended with preferences and Goals (ABA+G). Remote EHR integration of computable guidelines was via a microservice architecture based on HL7 FHIR and CDS Hooks. A prototype integration was developed to manage chronic obstructive pulmonary disease (COPD) with comorbid cardiovascular or chronic kidney diseases, and a mixed-methods evaluation was conducted with 20 simulated cases and five pulmonologists. Results: Pulmonologists agreed 97% of the time with the GOLD-based COPD symptom severity assessment assigned to each patient by the CDSS, and 98% of the time with one of the proposed COPD care plans. Comments were favourable on the principles of explainable argumentation; inclusion of additional co-morbidities was suggested in the future along with customisation of the level of explanation with expertise. Conclusion: An ontological model provided a flexible means of providing argumentation and explainable artificial intelligence for a long-term condition. Extension to other guidelines and multiple co-morbidities is needed to test the approach further.

Achieving descriptive accuracy in explanations via argumentation: The case of probabilistic classifiers.

The pursuit of trust in and fairness of AI systems in order to enable human-centric goals has been gathering pace of late, often supported by the use of explanations for the outputs of these systems. Several properties of explanations have been highlighted as critical for achieving trustworthy and fair AI systems, but one that has thus far been overlooked is that of descriptive accuracy (DA), i.e., that the explanation contents are in correspondence with the internal working of the explained system. Indeed, the violation of this core property would lead to the paradoxical situation of systems producing explanations which are not suitably related to how the system actually works: clearly this may hinder user trust. Further, if explanations violate DA then they can be deceitful, resulting in an unfair behavior toward the users. Crucial as the DA property appears to be, it has been somehow overlooked in the XAI literature to date. To address this problem, we consider the questions of formalizing DA and of analyzing its satisfaction by explanation methods. We provide formal definitions of naive, structural and dialectical DA, using the family of probabilistic classifiers as the context for our analysis. We evaluate the satisfaction of our given notions of DA by several explanation methods, amounting to two popular feature-attribution methods from the literature, variants thereof and a novel form of explanation that we propose. We conduct experiments with a varied selection of concrete probabilistic classifiers and highlight the importance, with a user study, of our most demanding notion of dialectical DA, which our novel method satisfies by design and others may violate. We thus demonstrate how DA could be a critical component in achieving trustworthy and fair systems, in line with the principles of human-centric AI.

Prevalence and determinants of hip pain in non-ambulatory cerebral palsy children: a retrospective cohort study.

BACKGROUND: Hip pain is common in cerebral palsy children, particularly at Gross-Motor Function Classification System level IV-V. It is associated to hip displacement and relates to the migration percentage. Recent literature suggested early reconstructive bone surgery, as the best approach to prevent hip luxation, then hip pain. Still, high rates of hip pain are reported. AIM: To investigate prevalence and determinants of hip pain in an Italian cerebral palsy sample. DESIGN: Single-center retrospective cohort study. SETTING: Inpatient and outpatient. POPULATION: Patients with spastic or dyskinetic cerebral palsy, Gross-Motor Function Classification System level IV or V, age 0-18. METHODS: A chart review was implemented to report hip pain, as a dichotomous variable (pain/no pain), age, sex, cerebral palsy subtype, Gross-Motor Function level, lumbar scoliosis, migration percentage, previous orthopedic surgery, or botulinum injections, oral or intrathecal baclofen, drug-resistant epilepsy, assistive devices for standing or walking. Descriptive statistics and a multivariate logistic stepwise regression were performed. RESULTS: A total of 504 subjects were included: 302 level V, 209 females, 432 spastics. The mean length of follow-up was 6 years. The overall prevalence of hip pain was 8.9% (6.3% were at level V) and of hip dislocation was 19% (15.9% were at level V). Just 39% of dislocated hips were painful. Children at spastic subtype and level V were predominantly affected. Botulinum and soft tissue surgery related to lower rates of hip pain, without statistical significance. Age (OR 1.19, 95%CI 1.14-1.25, P value 0.000), sex (OR 1.72, 95%CI 1.18-2.52, P value 0.005), migration percentage (OR 1.02, 95%CI 1.02-1.03, P value 0.000) and lumbar scoliosis (OR 1.32, 95%CI 0.86-2.01, P value 0.200) resulted significant independent determinants of hip pain. CONCLUSIONS: Hip pain relates with the migration percentage, but not all dislocated hips become painful. Hip pain may be transient and requires a targeted and individualized approach. Children at spastic subtype and level V were predominantly affected. Age and sex are confirmed as determinants. Specific validated measures are to be implemented to assess hip pain. CLINICAL REHABILITATION IMPACT: Considering severe non-ambulatory cerebral palsy patients, pain and quality of life should be considered as outcomes, in the management of hip luxation.

Surgical treatment of arteriovenous malformations of the buttock.

For symptomatic buttock arteriovenous malformations (AVMs), embolization techniques and surgical resection have been suggested as treatment options. Our aim was to evaluate the feasibility and long-term results after a single surgical resection. Twelve patients had undergone surgical resection without preoperative embolization. Of the 12 patients, 11 had had incomplete procedures, 9 of whom had undergone arterial embolization 1 to 3 years previously. All the patients were symptomatic. Computed tomography scans showed AVMs located in the cellular spaces with preservation of the gluteal muscle. The median follow-up time was 80 months. On the last follow-up computed tomography scan, 67% had had no residual AVM. The use of preoperative embolization, especially with nonresorbable embolic material (Onyx; Medtronic, Dublin, Ireland), makes AVM resection and imaging follow-up more difficult because of artifacts and should be avoided.

Hip subluxation in Italian cerebral palsy children and its determinants: a retrospective cohort study.

The study's aim was two-fold: to describe the trend of hip subluxation in the largest sample of Italian nonambulatory cerebral palsy (CP) children ever published; to investigate its determinants. This single-centre retrospective cohort study included patients with spastic or dyskinetic CP, Gross Motor Function Classification System (GMFCS) level IV or V, age 0-18 years, having been referred to our unit before March 2020. The hip subluxation was measured by means of the migration percentage (MP). Other data were gathered such as sex, CP subtype, GMFCS level, presence of drug-resistant epilepsy, age, use of walkers with weight relief or standing devices, previous botulinum injection or hip surgery, oral or intrathecal baclofen and hip pain. Multiple linear stepwise regression was performed and descriptive statistics are provided. Spastic CP had MP maximum increase in early ages, with GMFCS level V values persistently higher than level IV. The dyskinetic subtype showed a slower increase of the MP, with GMFCS level IV presenting similar or higher values, compared to level V. Age, CP severity and spastic subtype are the main determinants. The stepwise multiple regression analysis demonstrated that weight relief walking and standing assistive devices, combined with botulinum contributed to reduce the MP progression. Dyskinetic CP showed overall lower MP values and a more variable behaviour relative to age and GMFCS level, compared to the spastic subtype. Standing and walking assistive devices, with partial or total weight relief, combined with individually targeted botulinum injections, should be considered in the management of bilateral nonambulatory CP patients, to prevent hip subluxation or its recurrence after surgery.

A retrospective cohort study about hip luxation in non-ambulatory cerebral palsy patients: The point of no return.

Purpose: The migration percentage is a widely used criterion for surgery in displaced hips. Literature suggests that no hip can spontaneously improve if the migration percentage exceeds 45%, in a mixed population of cerebral palsy children. The aim of the present article was to identify the point of no return of the migration percentage in a selected sample of non-ambulatory cerebral palsy children, being the most exposed to hip luxation. Methods: This single-center retrospective cohort study included patients with spastic or dyskinetic cerebral palsy, Gross Motor Function Classification System level IV or V, age 0-18, having at least three pelvic radiographies, excluding radiographies relative to hips having previously undergone surgery. The following information was collected: sex, cerebral palsy subtype, Gross Motor Function Classification System level, presence of drug-resistant epilepsy, migration percentage, age at assessment, use of walking or standing assistive devices, previous botulinum injection, oral or intrathecal baclofen, and hip pain. Data were analyzed at the level of the individual hips. Descriptive statistics were presented. Receiver operating characteristic curve analysis was conducted to investigate which value of the migration percentage could be adopted as the "point of no return": that is, the cutoff value beyond which no migration percentage reduction, by more than 5%, could be expected. Results: The optimal cutoff value was identified as migration percentage ≥50%, with a sensitivity of 84.5% and a specificity of 100% (p-value <0.001). Conclusion: Based on the present study, migration percentage ≥50% is the "point of no return" for Gross Motor Function Classification System IV-V cerebral palsy patients, representing the cutoff value beyond which no spontaneous cerebral palsy reduction may be expected, unless addressing surgery. Level of evidence: level II-retrospective study.

Surgical treatment of vulvar venous malformations without preoperative sclerotherapy: Outcomes of 18 patients.

BACKGROUND: Our aim was to evaluate the benefit of surgical resection of the venous malformations (VMs) of the external female genitalia. METHODS: Over the period of 2009-2019, 18 consecutive females underwent surgical resection for vulvar VM. Evaluations included preoperative Doppler ultrasound, MRI, and pre-and postoperative photographic imaging. The main outcomes were: residual pain, cosmetic distortion, residual VM, and quality of life. RESULTS: Over a 10 year periods, 18 females, mean age 35 years (range 9-71) were included in this study. All patients were symptomatic: 16 had intermittent pain or discomfort, 1 had bleeding and 2 requested cosmetic treatment. Of these cases, there were 5 isolated vulvar VM, 12 associated VM: 3 of the clitoral hood, 3 troncular pelvic vein insufficiency and 12 of the lower limb. Eight patients had undergone previous procedures: 2 sclerotherapy treatments (1-3 sessions), 4 partial surgical resections. There were 18 single resections in the vulva (7 focal, 11 complete), 2 partial resections in clitoral hood and 2 had resection of a VM in the perirtoneovaginal canal at the same time. The mean follow-up was 42. 9 months (range 6-120). Two patients were lost to followup at 6 months. For all patients, elimination of pain and soft tissue redundancy was achieved. Two patients had persistent discomfort and 2 requested cosmetic treatment. CONCLUSION: Surgical resection of vulvar VM can be the best approach with few postoperative complications, good functional and cosmetic results. Appropriate preoperative evaluation is required to identify isolated VM or VM associated with ovarian vein or internal iliac vein insufficiency requiring to be treated before surgery.

Surgical Treatment of Venous Malformations of the Knee: Treatment Outcomes for 35 Patients.

BACKGROUND: The place of open surgery in venous malformations (VMs) of knee joint is still discussed. The aim of this study was to evaluate the benefits of surgery in terms of pain, function, and quality of life. DESIGN: This was a retrospective observational study. METHODS: Thirty-five consecutive young patients undergoing surgery for VMs of the knee between 2011 and 2014 were included. Data collection was performed using a prospective database by reviewing patient records. Pain, mobility of the joint, residual VMs as seen by magnetic resonance imaging, and quality of life were the main outcome endpoints for this study. RESULTS: Thirty-five patients (22 females and 13 males, with a median age of 15 y, range of 5 to 20 y) were included. Twenty-seven VMs were localized in and around the knee joint, of which 8 were extensive. Indication for surgery was intermittent or permanent pain.Details of the surgical excisions of the VM are as follows: suprapatellar area in 57%, suprapatellar and infrapatellar area in 29%, infrapatellar area in 2%, limited in the femoropatellar area in 12%. Partial resection of a vastus muscle or patellar retinaculum was necessary for 19 patients.The median duration of the surgery was 4 hours (range: 2 to 7 h). The median hospital stay was 5 days; full-time physiotherapy was systematic for 2 to 3 weeks.After 6 months, 74% had no longer pain, the mobility of the joint was normal for 60%. Of the 33 patients who had a magnetic resonance imaging at 6 months, 86% (28 patients) had no residual intra-articular VMs, and 14% had focal residual VMs.At the last follow-up, 77% (27 patients) had no longer pain, 23% experienced intermittent pain with unusual activities, and 91% (32 patients) had normal mobility of the joint. Quality of life was increased significantly due to the pronounced impact on pain. CONCLUSION: Extensive surgical excision of knee VMs is a safe procedure with good outcomes in terms of pain, function, and quality of life. LEVEL OF EVIDENCE: Level II-retrospective observational monocentric study.

Unexpected granular cell tumor in abdominal wall: case report and literature review.

Granular cell tumors (GCTs) are uncommon benign neoplasms deriving from Schwann cells of the peripheral nerve fibers. Although these tumors can be found anywhere in the body, the most frequent site is the tongue, followed by the chest wall and the arm. The abdominal wall is an extremely rare site for GCTs. These tumors are generally asymptomatic and have a slow growth rate. Today, thanks to their immunoreactivity to S-100 and CD68, the differential diagnosis is more straightforward than in the past. We report on a young patient affected by a GCT located in the upper third of the right rectus abdominis muscle. En bloc excision through a diamond-shaped skin incision allowed us to make a correct histological diagnosis, which was confirmed by the immunohistochemical findings. GCT, which is very rare in abdominal wall muscles, should be considered in the differential diagnosis, and surgical excision is the treatment of choice.