Artificial Neural Networks and risk stratification models in Emergency Departments: The policy maker's perspective

Translational artificial intelligence-led optimization and realization of estimated discharge with a supportive weekend interprofessional flow team (TAILORED-SWIFT)

Weekend discharges occur less frequently than discharges on weekdays, contributing to hospital congestion. Artificial intelligence algorithms have previously been derived to predict which patients are nearing discharge based upon ward round notes. In this implementation study, such an artificial intelligence algorithm was coupled with a multidisciplinary discharge facilitation team on weekend shifts. This approach was implemented in a tertiary hospital, and then compared to a historical cohort from the same time the previous year. There were 3990 patients included in the study. There was a significant increase in the proportion of inpatients who received weekend discharges in the intervention group compared to the control group (median 18%, IQR 18-20%, vs median 14%, IQR 12% to 17%, P = 0.031). There was a corresponding higher absolute number of weekend discharges during the intervention period compared to the control period (P = 0.025). The studied intervention was associated with an increase in weekend discharges and economic analyses support this approach as being cost-effective. Further studies are required to examine the generalizability of this approach to other centers.

Neural networks and hospital length of stay: an application to support healthcare management with national benchmarks and thresholds

Background

The problem of correct inpatient scheduling is extremely significant for healthcare management. Extended length of stay can have negative effects on the supply of healthcare treatments, reducing patient accessibility and creating missed opportunities to increase hospital revenues by means of other treatments and additional hospitalizations.

Methods

Adopting available national reference values and focusing on a Department of Internal and Emergency Medicine located in the North-West of Italy, this work assesses prediction models of hospitalizations with length of stay longer than the selected benchmarks and thresholds. The prediction models investigated in this case study are based on Artificial Neural Networks and examine risk factors for prolonged hospitalizations in 2018. With respect current alternative approaches (e.g., logistic models), Artificial Neural Networks give the opportunity to identify whether the model will maximize specificity or sensitivity.

Results

Our sample includes administrative data extracted from the hospital database, collecting information on more than 16,000 hospitalizations between January 2018 and December 2019. Considering the overall department in 2018, 40% of the hospitalizations lasted more than the national average, and almost 3.74% were outliers (i.e., they lasted more than the threshold). According to our results, the adoption of the prediction models in 2019 could reduce the average length of stay by up to 2 days, guaranteeing more than 2000 additional hospitalizations in a year.

Conclusions

The proposed models might represent an effective tool for administrators and medical professionals to predict the outcome of hospital admission and design interventions to improve hospital efficiency and effectiveness.

Rise of Clinical Studies in the Field of Machine Learning: A Review of Data Registered in ClinicalTrials.gov

Although advances in machine-learning healthcare applications promise great potential for innovative medical care, few data are available on the translational status of these new technologies. We aimed to provide a comprehensive characterization of the development and status quo of clinical studies in the field of machine learning. For this purpose, we performed a registry-based analysis of machine-learning-related studies that were published and first available in the ClinicalTrials.gov database until 2020, using the database’s study classification. In total, n = 358 eligible studies could be included in the analysis. Of these, 82% were initiated by academic institutions/university (hospitals) and 18% by industry sponsors. A total of 96% were national and 4% international. About half of the studies (47%) had at least one recruiting location in a country in North America, followed by Europe (37%) and Asia (15%). Most of the studies reported were initiated in the medical field of imaging (12%), followed by cardiology, psychiatry, anesthesia/intensive care medicine (all 11%) and neurology (10%). Although the majority of the clinical studies were still initiated in an academic research context, the first industry-financed projects on machine-learning-based algorithms are becoming visible. The number of clinical studies with machine-learning-related applications and the variety of medical challenges addressed serve to indicate their increasing importance in future clinical care. Finally, they also set a time frame for the adjustment of medical device-related regulation and governance.

Artificial neural networks and risk stratification in emergency departments

Emergency departments are characterized by the need for quick diagnosis under pressure. To select the most appropriate treatment, a series of rules to support decision-making has been offered by scientific societies. The effectiveness of these rules affects the appropriateness of treatment and the hospitalization of patients. Analyzing a sample of 1844 patients and focusing on the decision to hospitalize a patient after a syncope event to prevent severe short-term outcomes, this work proposes a new algorithm based on neural networks. Artificial neural networks are a non-parametric technique with the well-known ability to generalize behaviors, and they can thus predict severe short-term outcomes with pre-selected levels of sensitivity and specificity. This innovative technique can outperform the traditional models, since it does not require a specific functional form, i.e., the data are not supposed to be distributed following a specific design. Based on our results, the innovative model can predict hospitalization with a sensitivity of 100% and a specificity of 79%, significantly increasing the appropriateness of medical treatment and, as a result, hospital efficiency. According to Garson's Indexes, the most significant variables are exertion, the absence of symptoms, and the patient's gender. On the contrary, cardio-vascular history, hypertension, and age have the lowest impact on the determination of the subject's health status. The main application of this new technology is the adoption of smart solutions (e.g., a mobile app) to customize the stratification of patients admitted to emergency departments (ED)s after a syncope event. Indeed, the adoption of these smart solutions gives the opportunity to customize risk stratification according to the specific clinical case (i.e., the patient's health status) and the physician's decision-making process (i.e., the desired levels of sensitivity and specificity). Moreover, a decision-making process based on these smart solutions might ensure a more effective use of available resources, improving the management of syncope patients and reducing the cost of inappropriate treatment and hospitalization.

The private healthcare market and the sustainability of an innovative community nurses programme based on social entrepreneurship - CoNSENSo project

Background

CoNSENSo is a project funded by the European Union, which is aimed at developing an innovative care model based on community nurses to support active ageing in mountain areas. The planned sustainability of this innovative approach relies on social entrepreneurship on the healthcare market, and this work highlights the necessary conditions for the successful implementation of these entrepreneurial initiatives.

Methods

Considering municipalities in the Piedmont Region and those aged 65 or older as target population, the authors propose several negative binomial regression models to estimate the effectiveness of current private healthcare services in supporting the active aging process. Such effectiveness may represent the ex-ante (positive) reputation of these new social entrepreneurial initiatives on the market.

Results

According to our results, the private supply of healthcare services can effectively support the aging process. Indeed, given that the other predictor variables in the model are held constant, there are statistically significant negative relations between the number of hip fractures and the private supply of healthcare services by dental practitioners and psychologists (p-value < 0.05), as well as the private supply of opportunities for social interaction by coffee bars (p-value < 0.05).

Conclusions

The authors expect a favourable environment for the entrepreneurial initiatives of community nurses in mountain areas. Accordingly, policy makers cannot reject the hypothesis that the goals reached by the CoNSENSo project may be maintained for the sake of the future generations, avoiding its collapse as soon as public funding shifts to new programmes.

The Economic Impact of Clinical Research in an Italian Public Hospital: The Malignant Pleural Mesothelioma Case Study

Background: The current economic constraints cause hospital management to use the available public resources as rationally as possible. At the same time, there is the necessity to improve current scientific knowledge. This is even more relevant in the case of patients with malignant pleural mesothelioma (MPM), given the severity of the disease, its dismal prognosis, and the cost of chemotherapy drugs. This work aims to evaluate the standard cost of patients with MPM, supporting physicians in their decision-making process in relation to budget constraints, as well as policy-makers with respect research policy.

Methods: The authors conducted a retrospective cost analysis on all the patients with MPM who were first admitted to a reference hospital specialized in MPM care between 2014 and 2015, collecting data on their diagnostic pathways and active treatments, as well as on the related official fees for each procedure. Then, using a multiple regression model, we estimated the overall expected cost of a patient with MPM treated in our hospital, to be born by the Regional Healthcare System based on the chosen clinical pathway.

Results: According to results, the economic impact of caring for a patient with MPM is mostly related to the selected active treatments, with drug and hospitalization costs as main drivers. Our analysis suggests that the expected reimbursed fee to care for a patient with MPM is equal to € 18 214.99, with chemotherapy and monitoring costs equal to € 12 861.43 and hospitalization cost equal to € 5353.55. This cost decreases to € 320.18 in the case of enrollment in an experimental trial of first-line treatment. In the other cases (second-line or third-line trials), the expected cost borne by the healthcare system for treating patients grows exponentially (€ 40,124.18 and € 59 839.94, respectively).

Conclusion: Experimental trials might be a solution to decrease the economic burden for the public healthcare system only in the case of first-line treatments, where the cost of chemotherapy is relevant. Nevertheless, policy-makers have to accept the sharing of this economic burden between society and the pharmaceutical industry to broaden the current scientific knowledge.

The sustainability of a community nurses programme aimed at supporting active ageing in mountain areas

BACKGROUND

Community Nurse Supporting Elderly iN a changing SOciety is a project funded by the European Union, which is aimed at developing an innovative care model based on community nurses to support active ageing in mountain areas. The planned sustainability of this innovative approach relies on social entrepreneurship, and this work highlights the necessary conditions for the existence of these entrepreneurial initiatives on the market, with community nurses' services purchased by the public health care system.

METHODS

The authors propose a sustainability framework for this project based on three relevant dimensions (ie, health, organisation, and context), highlighting the necessary conditions for continued provision of health services beyond project conclusion. Then, considering the Piedmont Region and those aged 65 or older as target population, health outcomes are analysed, proposing a break-even analysis to calculate expected levels.

RESULTS

According to our results, in order to care for 191 977 elderly people for 3 years, a successful pro-active approach is needed to prevent 1657 falls with hip fracture, reducing the prevalence of this adverse outcome by 36%. These are the expected health outcome levels for the existence of a social market, which can be achieved through the successful involvement of local public health organisations and stakeholders.

CONCLUSIONS

Policy makers need clear information on the economic impact of extending this new intervention to the whole target population and on the required preconditions for its financial sustainability in terms of health outcomes. However, a participatory process involving all relevant local stakeholders and organisations is crucial to extend current achievements beyond project conclusion.

Neural networks as a tool to predict syncope risk in the Emergency Department

Aims

There is no universally accepted tool for the risk stratification of syncope patients in the Emergency Department. The aim of this study was to investigate the short-term predictive accuracy of an artificial neural network (ANN) in stratifying the risk in this patient group.

Methods and results

We analysed individual level data from three prospective studies, with a cumulative sample size of 1844 subjects. Each dataset was reanalysed to reduce the heterogeneity among studies defining abnormal electrocardiogram (ECG) and serious outcomes according to a previous consensus. Ten variables from patient history, ECG, and the circumstances of syncope were used to train and test the neural network. Given the exploratory nature of this work, we adopted two approaches to train and validate the tool. One approach used 4/5 of the data for the training set and 1/5 for the validation set, and the other approach used 9/10 for the training set and 1/10 for the validation set. The sensitivity, specificity, and area under the receiver operating characteristic curve of ANNs in identifying short-term adverse events after syncope were 95% [95% confidence interval (CI) 80-98%], 67% (95% CI 62-72%), 0.69 with the 1/5 approach and 100% (95% CI 84-100%), 79% (95% CI 72-85%), 0.78 with the 1/10 approach.

Conclusion

The results of our study suggest that ANNs are effective in predicting the short-term risk of patients with syncope. Prospective studies are needed in order to compare ANNs' predictive capability with existing rules and clinical judgment.