Artificial intelligence in healthcare: why not apply the medico-legal method starting with the Collingridge dilemma?

Technology has greatly influenced and radically changed human life, from communication to creativity and from productivity to entertainment. The authors, starting from considerations concerning the implementation of new technologies with a strong impact on people's everyday lives, take up Collingridge's dilemma and relate it to the application of AI in healthcare. Collingridge's dilemma is an ethical and epistemological problem concerning the relationship between technology and society which involves two approaches. The proactive approach and socio-technological experimentation taken into account in the dilemma are discussed, the former taking health technology assessment (HTA) processes as a reference and the latter the AI studies conducted so far. As a possible prevention of the critical issues raised, the use of the medico-legal method is proposed, which classically lies between the prevention of possible adverse events and the reconstruction of how these occurred.The authors believe that this methodology, adopted as a European guideline in the medico-legal field for the assessment of medical liability, can be adapted to AI applied to the healthcare scenario and used for the assessment of liability issues. The topic deserves further investigation and will certainly be taken into consideration as a possible key to future scenarios.

Partial Nephrectomy versus Percutaneous Cryoablation of Small Renal Cell Carcinomas: A Comparison of Adverse Events in a Prospective Multicenter Cohort Study.

PURPOSE: To assess and compare complications and readmissions after partial nephrectomy and percutaneous cryoablation of cT1 renal cell carcinoma (RCC). MATERIALS AND METHODS: Patients treated for cT1 RCC between 2019 and 2021 were prospectively and consecutively enrolled. Complications recorded within 30 and 90 days were graded according to the Clavien-Dindo classification, and percutaneous cryoablation was graded according to the Society of Interventional Radiology classification of adverse events. Major complications were defined as complications with a grade of ≥3 based on the Clavien-Dindo classification. Readmission within 30 days was recorded. RESULTS: The cohort included 86 partial nephrectomies and 104 cryoablations. The complication rate within 90 days was 23% after partial nephrectomy and cryoablation (P = .98), with major complication rates of 3% after partial nephrectomy and 10% after cryoablation (P = .15). The readmission rates were 14% and 11% after partial nephrectomy and cryoablation, respectively (P = .48). Double-J stents were associated with overall complications (odds ratio [OR], 9.88; 95% confidence interval [CI], 2.18-44.68; P = .003) and readmissions (OR, 5.39; 95% CI, 1.37-21.06; P = .015) after cryoablation. A high versus low radius-endophytic-nearness-anterior-location score (OR, 5.86; 95% CI, 1.08-31.81; P = .040) and endophytic location (OR, 7.70; 95% CI, 1.72-34.50; P = .008) were associated with a higher complication rate after cryoablation. The Charlson Comorbidity Index (CCI) was associated with major complications after partial nephrectomy (OR, 2.12; 95% CI, 1.05-4.30; P = .036). CONCLUSIONS: Partial nephrectomy and cryoablation are comparable regarding complications within 90 days after treatment. Tumor complexity and double-J stents were associated with complications after cryoablation, and a high CCI was associated with complications after partial nephrectomy.

Value assessment of artificial intelligence in medical imaging: a scoping review.

BACKGROUND: Artificial intelligence (AI) is seen as one of the major disrupting forces in the future healthcare system. However, the assessment of the value of these new technologies is still unclear, and no agreed international health technology assessment-based guideline exists. This study provides an overview of the available literature in the value assessment of AI in the field of medical imaging. METHODS: We performed a systematic scoping review of published studies between January 2016 and September 2020 using 10 databases (Medline, Scopus, ProQuest, Google Scholar, and six related databases of grey literature). Information about the context (country, clinical area, and type of study) and mentioned domains with specific outcomes and items were extracted. An existing domain classification, from a European assessment framework, was used as a point of departure, and extracted data were grouped into domains and content analysis of data was performed covering predetermined themes. RESULTS: Seventy-nine studies were included out of 5890 identified articles. An additional seven studies were identified by searching reference lists, and the analysis was performed on 86 included studies. Eleven domains were identified: (1) health problem and current use of technology, (2) technology aspects, (3) safety assessment, (4) clinical effectiveness, (5) economics, (6) ethical analysis, (7) organisational aspects, (8) patients and social aspects, (9) legal aspects, (10) development of AI algorithm, performance metrics and validation, and (11) other aspects. The frequency of mentioning a domain varied from 20 to 78% within the included papers. Only 15/86 studies were actual assessments of AI technologies. The majority of data were statements from reviews or papers voicing future needs or challenges of AI research, i.e. not actual outcomes of evaluations. CONCLUSIONS: This review regarding value assessment of AI in medical imaging yielded 86 studies including 11 identified domains. The domain classification based on European assessment framework proved useful and current analysis added one new domain. Included studies had a broad range of essential domains about addressing AI technologies highlighting the importance of domains related to legal and ethical aspects.

Model for ASsessing the value of Artificial Intelligence in medical imaging (MAS-AI).

OBJECTIVES: Artificial intelligence (AI) is seen as a major disrupting force in the future healthcare system. However, the assessment of the value of AI technologies is still unclear. Therefore, a multidisciplinary group of experts and patients developed a Model for ASsessing the value of AI (MAS-AI) in medical imaging. Medical imaging is chosen due to the maturity of AI in this area, ensuring a robust evidence-based model. METHODS: MAS-AI was developed in three phases. First, a literature review of existing guides, evaluations, and assessments of the value of AI in the field of medical imaging. Next, we interviewed leading researchers in AI in Denmark. The third phase consisted of two workshops where decision makers, patient organizations, and researchers discussed crucial topics for evaluating AI. The multidisciplinary team revised the model between workshops according to comments. RESULTS: The MAS-AI guideline consists of two steps covering nine domains and five process factors supporting the assessment. Step 1 contains a description of patients, how the AI model was developed, and initial ethical and legal considerations. In step 2, a multidisciplinary assessment of outcomes of the AI application is done for the five remaining domains: safety, clinical aspects, economics, organizational aspects, and patient aspects. CONCLUSIONS: We have developed an health technology assessment-based framework to support the introduction of AI technologies into healthcare in medical imaging. It is essential to ensure informed and valid decisions regarding the adoption of AI with a structured process and tool. MAS-AI can help support decision making and provide greater transparency for all parties.

Substantial reduction in the number of amputations among patients with diabetes: a cohort study over 16 years.

AIMS/HYPOTHESIS: The aim of this study was to describe the trends in rates of amputation among individuals with and without diabetes. METHODS: We studied amputation rates in the County of Funen (approximately 0.5 million residents) during the period 1996-2011. Amputations were identified from the hospital administrative system, diabetes status by linkage with the Danish National Diabetes Register, and mortality and population data by extraction from Statistics Denmark. Amputation rates were analysed using proportional hazard models. We analysed the incidence of the first amputation at each level as well as the incidence of further amputations, subdivided by level of amputation. RESULTS: During the period 1996-2011, a total of 2,832 amputations were performed, of which 1,285 were among patients with diabetes and 1,547 among individuals without diabetes. Relative to persons without diabetes, patients with diabetes had an HR for below-ankle amputations (BAAs) of 14.7 for men and 7.5 for women, and for from-ankle-to-knee amputations (BKAs) of 7.6 and 8.4 for men and women, respectively. For above-knee amputations (AKAs) the numbers were 4.0 for men and 3.7 for women. We found an annual reduction in BAA rates among patients with diabetes of 9.8%, and the annual reduction in BKA for patients with diabetes was 15.1%. CONCLUSIONS/INTERPRETATION: The amputation rate in patients with diabetes is still several-fold higher than in persons without diabetes, but the improvements in diabetes care in recent years have resulted in a steady decline in amputation rates among patients with diabetes from this Danish cohort.

Immersive Virtual Reality in Basic Point-of-Care Ultrasound Training: A Randomized Controlled Trial.

This study was aimed at comparing the learning efficacy of a traditional instructor-led lesson with that of a completely virtual, self-directed lesson in immersive virtual reality (IVR) in teaching basic point-of-care ultrasound (PoCUS) skills. We conducted a blinded, non-inferiority, parallel-group, randomized controlled trial in which final-year medical students were randomized to an instructor-led (n = 53) or IVR (n = 51) lesson. Participants' learning efficacy was evaluated by blinded assessors, who rated each participant's performance using the Objective Structured Assessment of Ultrasound Skills (OSAUS) assessment tool.The mean total scores for participants were 11.0 points (95% confidence interval: 9.8-12.2) for the instructor-led lesson and 10.3 points (95% confidence interval: 9.0-11.5) for the IVR lesson. No significant differences were observed between the groups with respect to total score (p = 0.36) or subgroup objectives of the OSAUS score (p = 0.34 for familiarity, p = 0.45 for image optimization, p = 0.96 for systematic approach and p = 0.07 for interpretation). Maintenance costs for both courses were estimated at 400 euros each. Startup costs for the instructor-led course were estimated 16 times higher than those for the IVR course. The learning efficacy of an instructor-led lesson on basic US did not differ significantly from that of a self-directed lesson in IVR, as assessed using the OSAUS. The results suggest that IVR could be an equivalent alternative to instructor-led lessons in future basic US courses, but further research is warranted to clarify the role of IVR in PoCUS courses.

Quality of life and complications after nephron-sparing treatment of renal cell carcinoma stage T1-a systematic review.

BACKGROUND: Despite the fact that nephron-sparing treatment is considered preferable from a surgical perspective patients' quality of life (QoL) following different types of nephron-sparing treatments remains unclear. PURPOSE: To investigate the quality of life and complications after nephron-sparing treatment of renal cell carcinomas of stage T1. MATERIALS AND METHODS: A systematic search of six databases was carried out. We included studies that reported the quality of life and complications in patients aged 18 years or older following nephron-sparing treatment of renal cell carcinoma stage T1. The quality assessment was performed using the Critical Appraisal Skills Programme (CASP) checklist for cohort studies and the CASP Randomized Controlled Trial Checklist. Data were analyzed using a narrative approach. RESULTS: Eight studies were included, six of which investigated QoL after partial nephrectomy and two after ablation therapies. Seven studies reported complications. Three studies reported higher QoL scores after partial nephrectomy compared to radical nephrectomy. Two studies showed that QoL increased or returned to baseline levels up to 12 months following partial nephrectomy. One study reported a gradual increase in QoL after radiofrequency ablation, and one study reported that all patients recovered to baseline QoL following cryoablation. Across studies, we found a complication rate up to 20% after partial nephrectomy and up to 12.5% after ablation therapy. CONCLUSIONS: The results of this systematic review suggest that nephron-sparing treatment appears to be superior or comparable to other treatment alternatives with regard to QoL outcomes. Additionally, based on the studies included in this review, partial nephrectomy appears to have a higher complication rate compared with ablation therapies. SYSTEMATIC REVIEW REGISTRATION: PROSPERO CRD42020155594.

Impact of Partial Nephrectomy and Percutaneous Cryoablation on Short-term Health-related Quality of Life-A Prospective Comparative Cohort Study.

Background: Partial nephrectomy (PN) is the gold standard for the treatment of stage cT1 renal cell carcinoma (RCC). However, the increasing incidence of RCC in the elderly population calls for alternative minimally invasive treatments to reduce the negative effects on patients' health-related quality of life (HRQoL) and subsequent healthy life expectancy. Objective: To assess and compare short-term HRQoL and self-reported health status after PN and percutaneous cryoablation (PCA) of patients treated for RCC stage cT1. Design setting and participants: Patients who underwent PN or PCA between 2019 and 2021 for RCC stage cT1 at two university hospitals in Denmark were assessed. The exclusion criteria included insufficient understanding of the Danish language, dementia, metastatic RCC, conversion to nephrectomy, and salvage procedures. Intervention: PN and PCA. Outcome measurements and statistical analysis: The European Organisation for the Research and Treatment of Cancer Quality of Life Questionnaire was distributed before treatment, and 14 and 90 d postoperatively. A linear mixed-effect model was used to analyze the changes from baseline to follow-up between PN and PCA treatment. Results and limitations: The cohort included 165 patients (PN: 79; PCA: 86). The completion rate was 96-98%. Patients receiving PCA were significantly older (median 69.1 vs 62.1 yr) and had lower scores on physical (p < 0.001) and role functioning (p = 0.009) than PN. A statistically significant change from baseline to 14 d was found for several HRQoL scales, which favored PCA over PN. However, the observed change was no longer significant at 90-d follow-up. Limitations include sample size and confounding by indication. Conclusions: This study found a significant difference between baseline and 14-d follow-up in several HRQoL and symptoms scales, favoring PCA over PN. However, no significant differences were observed in any HRQoL scales between PN and PCA of RCC stage cT1 from baseline to 90-d follow-up. Patient summary: Surgical removal and percutaneous cryoablation (freezing) of small tumors in the kidney had a similar impact on quality of life after 90 d.

A Deep Learning Algorithm for Radiographic Measurements of the Hip in Adults-A Reliability and Agreement Study.

Hip dysplasia (HD) is a frequent cause of hip pain in skeletally mature patients and may lead to osteoarthritis (OA). An accurate and early diagnosis may postpone, reduce or even prevent the onset of OA and ultimately hip arthroplasty at a young age. The overall aim of this study was to assess the reliability of an algorithm, designed to read pelvic anterior-posterior (AP) radiographs and to estimate the agreement between the algorithm and human readers for measuring (i) lateral center edge angle of Wiberg (LCEA) and (ii) Acetabular index angle (AIA). The algorithm was based on deep-learning models developed using a modified U-net architecture and ResNet 34. The newly developed algorithm was found to be highly reliable when identifying the anatomical landmarks used for measuring LCEA and AIA in pelvic radiographs, thus offering highly consistent measurement outputs. The study showed that manual identification of the same landmarks made by five specialist readers were subject to variance and the level of agreement between the algorithm and human readers was consequently poor with mean measured differences from 0.37 to 9.56° for right LCEA measurements. The algorithm displayed the highest agreement with the senior orthopedic surgeon. With further development, the algorithm may be a good alternative to humans when screening for HD.

The effect of forearm rotation on radiographic measurements of the wrist: an experimental study using radiostereometric analyses on cadavers.

BACKGROUND: Although dorsal/palmar tilt, radial inclination (RI), and ulnar variance (UV) are measurements commonly performed in wrist radiographs, the impact of forearm rotation on those measurements during the radiographic procedure is uncertain. Our aim was to determine the impact of supination and pronation on the reliability of measurements of tilt, RI, and UV. METHODS: Tantalum markers were inserted into the distal radius of 21 unfractured cadaver forearms. The forearms were radiographed in different degrees of supination and pronation. The exact degree of rotation was calculated with radiostereometric analyses. Tilt, RI, and UV were measured by two independent readers in a random and anonymised fashion. Association between forearm rotation and radiographic measurements was examined using linear regression. RESULTS: Forearm rotation significantly impacted the radiographically measured tilt. One degree of supination and pronation respectively increased and decreased palmar tilt with 0.68° and 0.44°, observers 1 and 2, respectively. As opposed to observer 1, observer 2 found that RI was significantly impacted by rotation with a slope of 0.08. Ulnar variance was not significantly impacted by rotation with linear regression slopes of 0.01° (95% confidence interval [CI] - 0.02-0.05, p = 0.490) and 0.02° (95% CI - 0.02-0.07; p = 0.288), observer 1 and observer 2, respectively. CONCLUSION: In unfractured forearms, the radiographically measured tilt was significantly affected by rotation. Palmar tilt increased with supination and decreased with pronation. Rotation significantly affected radial inclination, although of a magnitude that is probably not clinically relevant. No significant impact on UV was found.

Dorsal Tilt of the Distal Radius Fracture Changes With Forearm Rotation When Measured on Radiographs.

Purpose: This study examined the impact of pronation and supination on the reliability of the radiographically measured values of dorsal tilt, radial inclination (RI), and ulnar variance (UV) in cadaveric forearms with artificially created distal radius fractures. Methods: We prepared 21 human cadaveric forearms (11 right and 10 left) for radiostereometric analysis (RSA) by insertion of tantalum markers. Distal radius fractures were created midway between the marker segments. Radiographs and RSA images were taken at different degrees of supination and pronation. The precise degree of forearm rotation was calculated using RSA software. Two observers (H.B.T. and T.T.) independently measured tilt, RI, and UV on all radiographs in a blinded and randomized fashion. Univariate linear regression analyses were used to determine the relationship between forearm rotation and the measured radiographic values. Results: The radiographically measured value of tilt was significantly impacted by forearm rotation. Supinating or pronating the forearm by 10° decreased and increased, respectively, the radiographic value of dorsal tilt by approximately 3°. Conclusions: This study showed that the positioning of the fractured forearm during the radiographic procedure significantly impacted subsequent radiographic measurements of tilt. Dorsal tilt measurements increased (ie, fracture displacement measured more dorsal) with pronation and decreased (ie, fracture displacement measured more toward neutral, with less dorsal tilt) with supination of the forearm. However, measurements of RI (p = 0.12 and p = 0.55 for observer 1 and 2) and UV (p = 0.34 and p = 0.17, observer 1 and 2) were not significantly impacted by rotation. Clinical relevance: Treatment of a distal radius fracture is, at least to some extent, based on radiographic quantification of fracture deformity. Therefore, unreliable measurements may adversely influence clinical decision making.

Complication Incidences and Treatment Outcomes of Endovascular Aneurysm Repair-A Single-Center Long-Time Follow-Up Study.

BACKGROUND: Few long-time follow-up studies describe all complications, treatment outcome of complications, and mortality in relation to endovascular aneurysm repair (EVAR). The purpose of this study was to evaluate the incidence and treatment outcome including mortality of radiological visible complications related to the EVAR procedure at a single center with up to 10 years' surveillance. MATERIALS AND METHODS: Patients treated with EVAR from March 2006 to March 2016 at a Danish university hospital, 421 in total, were included. Patient and aneurysm characteristics, follow-up, and secondary intervention data were collected from a national database and medical records. Follow-up computed tomography angiography and plain abdominal X-ray reports were reviewed for complications. Scans and X-rays with suspected complications were evaluated by an interventional radiologist. RESULTS: A total of 172 complications in 147 patients, mainly in the beginning of the follow-up period, were found; 35% had a least one complication. The main part of complications (62%) was type II endoleaks, followed by stent graft stenosis (11%), type I endoleaks (9%), and stent graft occlusion (7%). A total of 66 (38%) complications, observed in 55 patients, were treated with reintervention, of which 77% were treated with endovascular procedures and 23% with surgical treatment, that is, 13% of all studied patients had a complication that required a reintervention. The remaining 2 of the 3 complications were treated conservatively. We found no increased all-cause mortality in connection with having a complication including those requiring reintervention. CONCLUSION: We presented a 10-year single-center study of EVAR. Many patients treated with EVAR had a radiological visible complication, mainly in the beginning of the follow-up period. Only a smaller fraction required reintervention and having a reintervention-requiring complication was not connected to increased mortality.

A Randomized Controlled Trial Comparing Telemedical and Standard Outpatient Monitoring of Diabetic Foot Ulcers.

OBJECTIVE: The role of telemedical monitoring in diabetic foot ulcer care is still uncertain. Our aim was to compare telemedical and standard outpatient monitoring in the care of patients with diabetic foot ulcers in a randomized controlled trial. RESEARCH DESIGN AND METHODS: Of the 736 screened individuals with diabetic foot ulcers, 401 met the eligibility criteria and were randomized between October 2010 and November 2014. The per-protocol telemedical monitoring consisted of two consultations in the patient's own home and one consultation at the outpatient clinic. Standard practice consisted of three outpatient clinic visits. The three-visit cycle was repeated until study end point. The study end points were defined as complete ulcer healing, amputation, or death. RESULTS: One hundred ninety-three individuals were randomized to telemedical monitoring and 181 to standard care. Demographics were similar in both groups. A cause-specific Cox proportional hazards model showed no difference in individuals monitored through telemedicine regarding wound healing (hazard ratio 1.11 [95% CI 0.87, 1.42], P = 0.42) or amputation (0.87 [0.54, 1.42], P = 0.59). We found a higher mortality incidence in the telemedical monitoring group compared with the standard outpatient monitoring group (8.68 [6.93, 10.88], P = 0.0001). CONCLUSIONS: The findings of no significant difference regarding amputation and healing between telemedical and standard outpatient monitoring seem promising; however, for telemedical monitoring, a higher mortality throws into question the role of telemedicine in monitoring diabetic foot ulcers. Further studies are needed to investigate effects of telemedicine on mortality and other clinical outcomes and to identify patient subgroups that may have a poorer outcome through telemedical monitoring.