An innovative artificial intelligence-based method to compress complex models into explainable, model-agnostic and reduced decision support systems with application to healthcare (NEAR).

BACKGROUND AND OBJECTIVE: In everyday clinical practice, medical decision is currently based on clinical guidelines which are often static and rigid, and do not account for population variability, while individualized, patient-oriented decision and/or treatment are the paradigm change necessary to enter into the era of precision medicine. Most of the limitations of a guideline-based system could be overcome through the adoption of Clinical Decision Support Systems (CDSSs) based on Artificial Intelligence (AI) algorithms. However, the black-box nature of AI algorithms has hampered a large adoption of AI-based CDSSs in clinical practice. In this study, an innovative AI-based method to compress AI-based prediction models into explainable, model-agnostic, and reduced decision support systems (NEAR) with application to healthcare is presented and validated. METHODS: NEAR is based on the Shapley Additive Explanations framework and can be applied to complex input models to obtain the contributions of each input feature to the output. Technically, the simplified NEAR models approximate contributions from input features using a custom library and merge them to determine the final output. Finally, NEAR estimates the confidence error associated with the single input feature contributing to the final score, making the result more interpretable. Here, NEAR is evaluated on a clinical real-world use case, the mortality prediction in patients who experienced Acute Coronary Syndrome (ACS), applying three different Machine Learning/Deep Learning models as implementation examples. RESULTS: NEAR, when applied to the ACS use case, exhibits performances like the ones of the AI-based model from which it is derived, as in the case of the Adaptive Boosting classifier, whose Area Under the Curve is not statistically different from the NEAR one, even the model's simplification. Moreover, NEAR comes with intrinsic explainability and modularity, as it can be tested on the developed web application platform (https://neardashboard.pythonanywhere.com/). CONCLUSIONS: An explainable and reliable CDSS tailored to single-patient analysis has been developed. The proposed AI-based system has the potential to be used alongside the clinical guidelines currently employed in the medical setting making them more personalized and dynamic and assisting doctors in taking their everyday clinical decisions.

A Nurse-Led Model of Care with Telemonitoring to Manage Patients with Heart Failure in Primary Health Care: A Mixed-Method Feasibility Study.

Purpose: To determine whether a nurse-led care model with telemonitoring in primary care for patients with stable heart failure and their caregivers is feasible and acceptable. Patients and Methods: A mixed-methods feasibility study was conducted. Patients with stable heart failure and their caregivers were consecutively enrolled from March 2021 to April 2022. Participants were managed by nurses in a community health center through education and monitoring with a mobile app. The outcomes were feasibility outcomes, self-care outcomes, and qualitative acceptability and satisfaction. Quantitative and qualitative outcomes were linked to understanding how the model of care might benefit patients. Results: Twenty-six patients and nine of their caregivers were enrolled. Ten participants used the mobile app. Nineteen patients and eight caregivers were interviewed. Participants who improved their self-care appreciated the help in finding coping strategies, being close to the clinic, and feeling cared for. Participants with fewer improvements in self-care perceived the model of care as useless and were far from the centre. Participants decided to use the app mainly for usefulness prevision, and most of them were satisfied. Conclusion: The model of care was not successful in recruiting patients, and adjustments are needed to improve the recruitment strategy and to engage people who perceive the model of care as not useful or unable to use the app.

[The TeMP_cardio Model for the district management of heart failure patients: a feasibility study for the implementation of the family nurse]. / Il modello TeMP_Cardio per la gestione del paziente con scompenso cardiaco sul territorio: uno studio di fattibilità con l'introduzione dell'infermiere di famiglia e comunità.

. The TeMP_cardio Model for the district management of heart failure patients: a feasibility study for the implementation of the family nurse. INTRODUCTION: The need to enhance primary health nursing care and chronic disease management requires the development of complex and feasible models in local contexts. AIM: To test the feasibility of a complex care model, based on the introduction of the family and community nurse for patients with stable heart failure and their caregivers. METHODS: A pre-post six-month feasibility study was conducted in 2021-2022 in the Autonomous Province of Trento. Patients with stable heart failure able to access services were taken in charge with a multidisciplinary model, with the nurse as care manager, and technological and telemonitoring support. RESULTS: 26 patients out of 137 were included and nine of their caregivers. Ten participants accepted to use the mobile app. Twenty-three patients and seven caregivers completed the study, and the app was used regularly by five participants. Self-care skills improved in patients, mostly in self-monitoring, less in caregivers. One access to the emergency department for heart failure exacerbation was detected. CONCLUSIONS: The model was evaluated feasible to retain patients, with the need to build effective strategies for the recruitment, and the regular use of telemonitoring tools.

Introduction to the cognitive hospital.

The third industrial revolution has radically impacted the transformation of hospitals. Through the adoption of key digital technologies, hospitals have become more accessible, flexible, organised, responsive and able to deliver more personalised care. The digitalisation of patient health records, one of the most remarkable achievements to date in healthcare management, has enabled new opportunities, including the idea of hospitals evolving to become artificially intelligent. In parallel, the adoption of electronic and mobile internet technologies in hospitals has introduced new structural concepts, seeing a variety of terms blossom such as 'smart', 'intelligent', 'green' and 'liquid'. Now in the early fourth industrial revolution, driven by AI and internet-of-things technologies, this article unveils a new concept adapted to the upcoming era.

Implementation of tele visit healthcare services triggered by the COVID-19 emergency: the Trentino Province experience.

AIM: In response to the SARS-CoV-2 emergency, the Competence Centre on digital health 'TrentinoSalute4.0' has developed TreC_Televisita, a tele visit solution that meets the needs of the Trentino healthcare system and maintains high-quality patient-doctor interactions while respecting social distancing. This paper highlights how 'TreC_Televisita' was integrated into the Trentino healthcare system and its potential to become a structural and durable solution for the future local healthcare service provisioning. SUBJECT AND METHODS: This paper presents the multifactorial context that TreC_Televisita has faced for its implementation and the strategies adopted for its structural integration into the healthcare system. The analysis focuses on the main issues faced for the integration of the tele visits (e.g. privacy, payments) and how the context of TrentinoSalute4.0 permitted responding quickly to its implementation during the pandemic. It also describes how TreC_Televisita fits into the healthcare continuum from the organisational and technological standpoint, the end-user perspective and the barriers that could hamper the solution scalability. RESULTS: TreC_Televisita has demonstrated to be a technological solution that can be contextualised for different clinical domains beyond SARS-CoV-2. Moreover, it has shown its potential to scale up the solution beyond the COVID-19 emergency to the whole healthcare provisioning system in the long term. CONCLUSION: Being a positive experience in the first months of its implementation, the long-term goal is to transform TreC_Televisita into a structural pillar of the Trentino healthcare system, setting the bases for a sustainable, win-win situation for all the stakeholders involved in healthcare service provisioning.

Enabling teleophthalmology during the COVID-19 pandemic in the Province of Trento, Italy: Design and implementation of a mHealth solution.

BACKGROUND: Due to the many restrictions imposed during the COVID-19 emergency, the normal clinical activities have been stopped abruptly in view of limiting the circulation of the virus. The extraordinary containment measures have had a dramatic impact on the undertaking and follow-up of ophthalmic outpatients. OBJECTIVE: In order to guarantee proper monitoring and routine care, the Pediatric Ophthalmology equipe of Rovereto Hospital (North-East of Italy) supported by the Competence Center on Digital Health TrentinoSalute4.0, designed and implemented a digital platform, TreC Oculistica, enabling teleophthalmology. We report our innovative-albeit restricted-experience aiming at testing and maximizing the efficacy of remote ophthalmic and orthoptic visits. METHODS: A multidisciplinary team created the TreC Oculistica platform and defined a teleophthalmology protocol. The system consists of a clinician web interface and a patient mobile application. Clinicians can prescribe outpatients with the App and some preliminary measurements to be self-collected before the televisit. The App conveys the clinician's requests (i.e. measurements) and eases the share of the collected information in a secure digital environment, promoting a new health care workflow. RESULTS: Four clinicians took part in the testing phase (2 ophthalmologists and 2 orthoptists) and recruited 37 patients (mostly pediatric) in 3 months. Thanks to a continuous feedback between the testing and the technical implementation, it has been possible to identify pros and cons of the implemented functionalities, considering possible improvements. Digital solutions such as TreC Oculistica advance the digitalization of the Italian health care system, promoting a structured and effective reorganization of the workload supported by digital systems. CONCLUSIONS: The study tested an innovative digital solution in the teleophthalmology context and represented the first experience within the Italian healthcare system. This solution opens up new possibilities and scenarios that can be effective not only during the pandemic, but also in the traditional management of public health services.

Use of eHealth Platforms and Apps to Support Monitoring and Management of Home-Quarantined Patients With COVID-19 in the Province of Trento, Italy: App Development and Implementation.

BACKGROUND: Italy was the first country to largely experience the COVID-19 epidemic among other Western countries during the so-called first wave of the COVID-19 pandemic. Proper management of an increasing number of home-quarantined individuals created a significant challenge for health care authorities and professionals. This was especially true when considering the importance of remote surveillance to detect signs of disease progression and consequently regulate access to hospitals and intensive care units on a priority basis. OBJECTIVE: In this paper, we report on an initiative promoted to cope with the first wave of the COVID-19 epidemic in the Spring/Summer of 2020, in the Autonomous Province of Trento, Italy. A purposefully built app named TreCovid19 was designed to provide dedicated health care staff with a ready-to-use tool for remotely monitoring patients with progressive symptoms of COVID-19, who were home-quarantined during the first wave of the epidemic, and to focus on those patients who, based on their self-reported clinical data, required a quick response from health care professionals. METHODS: TreCovid19 was rapidly developed to facilitate the monitoring of a selected number of home-quarantined patients with COVID-19 during the very first epidemic wave. The app was built on top of an existing eHealth platform, already in use by the local health authority to provide home care, with the following functionalities: (1) to securely collect and link demographic and clinical information related to the patients and (2) to provide a two-way communication between a multidisciplinary health care team and home-quarantined patients. The system supported patients to self-assess their condition and update the multidisciplinary team on their health status. The system was used between March and June 2020 in the province of Trento. RESULTS: A dedicated multidisciplinary group of health care professionals adopted the platform over a period of approximately 3 months (from March-end to June 2020) to monitor a total of 170 patients with confirmed COVID-19 during home quarantine. All patients used the system until the end of the initiative. The TreCovid19 system has provided useful insights of possible viability and impact of a technological-organizational asset to manage a potentially critical workload for the health care staff involved in the periodic monitoring of a relevant number of quarantined patients, notwithstanding its limitations given the rapid implementation of the whole initiative. CONCLUSIONS: The technological and organizational model adopted in response to the COVID-19 pandemic was developed and finalized in a relatively short period during the initial few weeks of the epidemic. The system successfully supported the health care staff involved in the periodic monitoring of an increasing number of home-quarantined patients and provided valuable data in terms of disease surveillance.

Blood Lactate Concentration Prediction in Critical Care.

Blood lactate concentration is a reliable risk indicator of deterioration in critical care requiring frequent blood sampling. However, lactate measurement is an invasive procedure that can increase risk of infections. Yet there is no clinical consensus on the frequency of measurements. In response we investigate whether machine learning algorithms can be used to predict blood lactate concentration from ICU health records. We evaluate the performance of different prediction algorithms using a multi-centre critical care dataset containing 13,464 patients. Furthermore, we analyse impact of missing value handling methods in prediction performance for each algorithm. Our experimental analysis show promising results, establishing a baseline for further investigation into this problem.

Short-term outcome associated with remote evaluation (telecardiology) of patients with cardiovascular diseases during the COVID-19 pandemic.

INTRODUCTION: During the recent COVID-19 outbreak, Italian health authorities mandated to replace in-person outpatient evaluations with remote evaluations. METHODS: From March 16th 2020 to April 22th 2020, all outpatients scheduled for in-person cardiac evaluations were instead evaluated by phone. We aimed to report the short-term follow-up of 345 patients evaluated remotely and to compare it with a cohort of patients evaluated in-person during the same period in 2019. RESULTS: During a mean follow-up of 54 ± 11 days, a significantly higher proportion of patients evaluated in-person in 2019 visited the emergency department or died for any cause (39/391, 10% versus 13/345 3.7%, p = 0.001) and visited the emergency department for cardiovascular causes (19/391, 4.9% versus 7/345, 2.0%, p = 0.04) compared to 2020. No cardiovascular death was recorded in the two periods. To an evaluation with a satisfaction questionnaire 49% of patients would like to continue using remote controls in addition to traditional ones. CONCLUSION: These findings may have important implications for the management of patients during the current COVID-19 pandemic because they suggest that remote cardiovascular evaluations may replace in-hospital visits for a limited period.

Impact of Cusp Repair on Reoperation Risk After the David Procedure.

BACKGROUND: We assessed whether additional cusp repair during valve-sparing aortic root replacement affects the echocardiographic mid-term results; a subgroup analysis among patients with bicuspid aortic valve (BAV) and tricuspid aortic valve (TAV) was performed. METHODS: Between June 2002 and May 2015, 157 consecutive patients underwent valve-sparing aortic root replacement with the David technique. Thirty patients (19%) had BAV. In 19 patients (12%), cusp motion or anatomic abnormalities contributed in determining aortic regurgitation requiring an additional cusp repair. Mean follow-up was 7 ± 3.4 years. RESULTS: The cumulative 1-, 5-, and 12-year survival rates were 98%, 94%, and 90%, respectively. Fourteen patients (9%) required aortic valve replacement during follow-up. In 2 patients the underlying cause was bacterial endocarditis. Freedom from aortic valve reoperation was 96% at 1 year, 92% at 5 years, and 89% at 12 years. Reoperation rate was significantly higher (p < 0.001) in patients who received leaflet repair compared with patients who did not, with a freedom from reoperation at 8 years of 58% versus 94%. Among patients with BAV, those who did not require cusp repair had a freedom from reoperation at 8 years of 94%, with a significant difference compared with patients who received cusp repair (p = 0.04). Cusp repair did not affect reoperation risk in patients with tricuspid aortic valve. CONCLUSIONS: Adjunctive cusp repair seems to affect the mid-term reoperation risk in patients with BAV and not in patients with tricuspid aortic valve. We recommend caution in using this technique in case of asymmetric BAV requiring cusp repair.

Conventional versus frozen elephant trunk surgery for extensive disease of the thoracic aorta.

OBJECTIVE: To compare early and mid-term outcomes after repair of extensive aneurysm of the thoracic aorta using the conventional elephant trunk or frozen elephant trunk (FET) procedures. METHODS: Fifty-seven patients with extensive thoracic aneurysmal disease were treated using elephant trunk (n = 36) or FET (n = 21) procedures. Patients with aortic dissection, descending thoracic aorta (DTA) diameter less than 40 mm, and thoracoabdominal aneurysms were excluded from the analysis, as were those who did not undergo antegrade selective cerebral perfusion during circulatory arrest. Short-term and mid-term outcomes were compared according to elephant trunk/FET surgical management. RESULTS: Preoperative and intraoperative variables were similar in the two groups, except for a higher incidence of female sex, coronary artery disease and associated procedures in elephant trunk patients. Hospital mortality (elephant trunk: 13.9% versus FET: 4.8%; P = 0.2), permanent neurologic dysfunction (elephant trunk: 5.7% versus FET: 9.5%; P = 0.4) and paraplegia (elephant trunk: 2.9% versus FET: 4.8%; P = 0.6) rates were similar in the two groups. Follow-up was 100% complete. In the elephant trunk group, 68.4% of patients did not undergo a second-stage procedure during follow-up for a variety of reasons. Of these patients, the DTA diameter was greater than 51 mm in 72.2% and two (6.7%) died due to aortic rupture while awaiting stage-two intervention. Endovascular second-stage procedures were successfully performed in all FET patients with residual DTA aneurysmal disease (n = 3), whereas nine of 11 elephant trunk patients who returned for second-stage procedures required conventional surgical replacement through a lateral thoracotomy. Kaplan-Meier estimate of 4-year survival was 75.8 ±â€Š7.6 and 72.8 ±â€Š10.6 in elephant trunk and FET patients, respectively (log-rank P = 0.8). CONCLUSION: In patients with extensive aneurysmal disease of thoracic aorta, elephant trunk and FET procedures seem to be associated with similar satisfactory early and mid-term outcomes. The FET approach leads to single-stage treatment of all aortic disease in most patients, and facilitates endovascular second-stage treatment in patients with residual DTA disease. The elephant trunk staged-approach appears to leave a considerable percentage of patients at risk for adverse aortic events.

Aortic arch reoperation in a single centre: early and late results in 57 consecutive patients.

OBJECTIVES: Aortic arch surgery following previous replacement of the ascending aorta has become more frequent due to the improved long-term survival of patients after the primary procedure and advances in the management of complex repeat cardiac procedures. We retrospectively analysed our results of redo aortic arch surgery. METHODS: Between January 1995 and December 2011, a total of 1022 patients underwent surgery on the ascending aorta with or without involvement of the aortic arch. Of these, 57 patients (5%) underwent reoperations involving the aortic arch. Indications for repeat aortic arch surgery included arch aneurysm in 50%, residual aortic dissection with aneurysmal formation in 38%, and graft infection in 9% of patients. One patient (1%) underwent previous heart transplantation. Nine patients (16%) had Marfan syndrome. RESULTS: Mean age was 55 ± 15 years and 23 patients (40%) were female. Logistic EuroSCORE-predicted risk of mortality was 38 ± 19%. The mean time interval between primary and redo surgery was 7.6 ± 7 years for aortic aneurysm and 4.4 ± 4 years in aortic dissection patients (P = 0.09). Total and partial arch replacements were performed in 52 and 47% of patients, respectively. Mean circulatory arrest time was 28.3 ± 22.1 min, with selective antegrade cerebral perfusion (SACP) utilized in 68% of patients. Overall incidence of stroke or transient neurological deficit was 15%. The 30-day mortality was 9% (n = 5). Multivariable logistic regression analysis revealed that previous type A aortic dissection was the only independent risk factor (OR 3.7, 95% CI 1.5-5.5, P = 0.01) for 30-day mortality. Mean survival was 5.5 ± 0.5 years, and estimated 5-year survival was 74.9 ± 0.6%. Survival of patients undergoing reoperation for residual aortic dissection was significantly worse compared with patients with aortic arch aneurysm (log-rank P = 0.016). CONCLUSIONS: Aortic arch reoperation, although a technically complex operation, can be performed with acceptable mortality and morbidity. Patients with aortic dissection during their previous operation required repeat surgery significantly earlier when compared with patients with arch aneurysms, and had worse long-term survival.

Isolated tricuspid valve surgery in patients with previous cardiac surgery.

OBJECTIVES: Few studies have been published in literature on outcomes of isolated tricuspid valve (TV) surgery when performed as a reoperation. Hence, we analyzed our early and midterm results of TV surgery in this unique group of patients. METHODS: We performed a retrospective analysis of 82 consecutive patients who underwent isolated TV surgery as a reoperation at our institution between 1997 and 2010. Symptomatic TV regurgitation (84.2%), acute endocarditis (14.6%), and valve thrombosis after TV repair (1.2%) were the indications for surgery. A minimally invasive access through a right anterolateral thoracotomy was the preferred approach in 60% of the patients. Previous cardiac operations included mitral, aortic, and TV surgery in 60%, 29%, and 27% and coronary bypass surgery in 18%, usually performed as combined procedures. Elective surgery was performed in 67.1% of the patients. Mean patient age was 64.1 ± 11.9 years, 28% being male with an average logistic EuroSCORE of 16.4% ± 14.3%. Follow-up was 96% complete, with a mean duration of 2.6 ± 2.4 years. RESULTS: Overall thirty-day mortality was 14.6%; for patients without and with endocarditis, it was 12.9% and 25%. Thirty-day mortality for patients undergoing elective surgery was 4.0%. Overall 2-year survival was 63.0% ± 5.5%. The 2-year freedom from TV-related reoperation was 93.5% ± 3.3%. CONCLUSIONS: Postoperative results of isolated TV surgery as a reoperation are acceptable when performed electively but dismal in patients undergoing nonelective surgery. Thus, redo TV surgery, when indicated, should be performed sooner rather than later. Minimally invasive surgery through a right lateral minithoracotomy is a safe approach for patients with elective surgery.

Experience with the conventional and frozen elephant trunk techniques: a single-centre study.

OBJECTIVES: The treatment of patients with extensive thoracic aortic disease involving the arch and descending/thoracoabdominal aorta is often performed using an elephant trunk procedure. We retrospectively analysed our results comparing two different techniques: the conventional elephant trunk (cET) and the frozen elephant trunk (FET) operation. METHODS: Between January 2003 and December 2011, 171 consecutive patients underwent total aortic arch replacement with either a cET (n = 125) or FET (n = 46) technique. The mean age was 64 ± 13 years and was significantly higher in the FET group (P < 0.01). Acute Type A aortic dissection was the indication for surgery in 53.6% of cET and 17.4% of FET patients, and degenerative or atherosclerotic aneurysm accounted for 33.6% of cET and 58.7% of FET patients. The remaining patients were operated on for chronic Type A or acute or chronic Type B dissections with arch involvement. RESULTS: In-hospital mortality was 21.6 vs 8.7% for cET and FET patients, respectively (P = 0.1). Logistic regression analysis revealed Type A aortic dissection (odds ratio (OR) 3.1, P = 0.01) as the only independent predictor of hospital mortality. Stroke occurred in 16 vs 13% of cET vs FET patients (P = 0.4). Type A aortic dissection was an independent predictor of stroke by multivariable analysis (OR 2.6, P = 0.03), and axillary arterial cannulation was protective against stroke (OR 0.4, P = 0.04). The occurrence of new-onset paraplegia was significantly higher in the FET group (21.7 vs 4.0%, P < 0.001), and aortic repair with the FET technique was an independent predictor for paraplegia (OR 6.6, P = 0.001). Among patients receiving FET, a body core temperature during circulatory arrest of ≥ 28 °C in combination with a prolonged circulatory arrest time of >40 min was an independent predictor for permanent spinal cord injury (OR 5.0, 95% CI 1.1-20, P = 0.038). The estimated 1-, 3- and 5-year survival were 70 ± 4, 70 ± 4 and 68 ± 4% (cET) and 4 ± 7 and 60 ± 9, 40 ± 1% (FET), with mean survival time 5.2 ± 0.3 vs 3.8 ± 0.5 years (cET vs FET, log-rank P = 0.9). CONCLUSIONS: The FET procedure for extensive thoracic aortic disease is associated with an acceptable mortality rate, but with a higher incidence of perioperative spinal cord injury than cET. Arch replacement with a cET technique should be strongly considered in patients with expected prolonged circulatory arrest times, particularly if operated on under mild or moderate hypothermia. Axillary cannulation is associated with superior neurological outcomes and Type A acute aortic dissection is a risk factor for mortality and poor neurological outcomes in this patient population.

Posttraumatic stress disorder and depression in heart transplantation recipients: the relationship with outcome and adherence to medical treatment.

OBJECTIVE: There is growing evidence of the importance of psychiatric risk factors for predicting the outcome of heart transplantation (HT) recipients. The aim of our study was to investigate the role of major depression and posttraumatic stress disorder (PTSD) in the prediction of the outcome of HT in a consecutive sample of 107 recipients. METHOD: All subjects of the study underwent a structured diagnostic interview for assessing the presence of pretransplant and posttransplant major depression and transplantation-related PTSD 1 to 5 years after HT. The adherence to medical treatment was assessed some months after the structured interview. The medical outcome (acute rejections, cancer, mortality) was followed up for 8 years on average after the interview, using a prospective design. RESULTS: Estimated frequency of psychiatric diagnoses after HT was 12% for transplantation-related PTSD and 41% for major depression. The presence of an episode of major depression prior to HT is a significant independent risk factor for posttransplant malignancies. Age, posttransplant malignancies and poor adherence are significant predictors of mortality in the survival analyses. CONCLUSIONS: The present study highlights the importance of the assessment of psychosocial variables and psychiatric diagnoses before and after transplantation in HT recipients. Our findings have important clinical implications and require replication with larger samples.