Prediction of pharmaceutical and non-pharmaceutical expenditures associated with Diabetes Mellitus type II based on clinical risk.

OBJECTIVE: To assess the effectiveness of different machine learning models in estimating the pharmaceutical and non-pharmaceutical expenditures associated with Diabetes Mellitus type II diagnosis, based on the clinical risk index determined by the analysis of comorbidities. MATERIALS AND METHODS: In this cross-sectional study, we have used data from 11,028 anonymized records of patients admitted to a high-complexity hospital in Bogota, Colombia between 2017-2019 with a primary diagnosis of Diabetes. These cases were classified according to Charlson's comorbidity index in several risk categories. The main variables analyzed in this study are hospitalization costs (which include pharmaceutical and non-pharmaceutical expenditures), age, gender, length of stay, medicines and services consumed, and comorbidities assessed by the Charlson's index. The model's dependent variable is expenditure (composed of pharmaceutical and non-pharmaceutical expenditures). Based on these variables, different machine learning models (Multivariate linear regression, Lasso model, and Neural Networks) were used to estimate the pharmaceutical and non-pharmaceutical expenditures associated with the clinical risk classification. To evaluate the performance of these models, different metrics were used: Mean Absolute Percentage Error (MAPE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Coefficient of Determination (R2). RESULTS: The results indicate that the Neural Networks model performed better in terms of accuracy in predicting pharmaceutical and non-pharmaceutical expenditures considering the clinical risk based on Charlson's comorbidity index. A deeper understanding and experimentation with Neural Networks can improve these preliminary results, therefore we can also conclude that the main variables used and those that were proposed can be used as predictors for the medical expenditures of patients with diabetes type-II. CONCLUSIONS: With the increase of technology elements and tools, it is possible to build models that allow decision-makers in hospitals to improve the resource planning process given the accuracy obtained with the different models tested.

Length of Stay Prediction With Standardized Hospital Data From Acute and Emergency Care Using a Deep Neural Network.

OBJECTIVE: Length of stay (LOS) is an important metric for the organization and scheduling of care activities. This study sought to propose a LOS prediction method based on deep learning using widely available administrative data from acute and emergency care and compare it with other methods. PATIENTS AND METHODS: All admissions between January 1, 2011 and December 31, 2019, at 6 university hospitals of the Hospices Civils de Lyon metropolis were included, leading to a cohort of 1,140,100 stays of 515,199 patients. Data included demographics, primary and associated diagnoses, medical procedures, the medical unit, the admission type, socio-economic factors, and temporal information. A model based on embeddings and a Feed-Forward Neural Network (FFNN) was developed to provide fine-grained LOS predictions per hospitalization step. Performances were compared with random forest and logistic regression, with the accuracy, Cohen kappa, and a Bland-Altman plot, through a 5-fold cross-validation. RESULTS: The FFNN achieved an accuracy of 0.944 (CI: 0.937, 0.950) and a kappa of 0.943 (CI: 0.935, 0.950). For the same metrics, random forest yielded 0.574 (CI: 0.573, 0.575) and 0.602 (CI: 0.601, 0.603), respectively, and 0.352 (CI: 0.346, 0.358) and 0.414 (CI: 0.408, 0.422) for the logistic regression. The FFNN had a limit of agreement ranging from -2.73 to 2.67, which was better than random forest (-6.72 to 6.83) or logistic regression (-7.60 to 9.20). CONCLUSION: The FFNN was better at predicting LOS than random forest or logistic regression. Implementing the FFNN model for routine acute care could be useful for improving the quality of patients' care.

Adjuvant therapeutic strategy decision support for an elderly population with localized breast cancer: A monocentric cohort retrospective study.

Guidelines for the management of elderly patients with early breast cancer are scarce. Additional adjuvant systemic treatment to surgery for early breast cancer in elderly populations is challenged by increasing comorbidities with age. In non-metastatic settings, treatment decisions are often made under considerable uncertainty; this commonly leads to undertreatment and, consequently, poorer outcomes. This study aimed to develop a decision support tool that can help to identify candidate adjuvant post-surgery treatment schemes for elderly breast cancer patients based on tumor and patient characteristics. Our approach was to generate predictions of patient outcomes for different courses of action; these predictions can, in turn, be used to inform clinical decisions for new patients. We used a cohort of elderly patients (≥ 70 years) who underwent surgery with curative intent for early breast cancer to train the models. We tested seven classification algorithms using 5-fold cross-validation, with 80% of the data being randomly selected for training and the remaining 20% for testing. We assessed model performance using accuracy, precision, recall, F1-score, and AUC score. We used an autoencoder to perform dimensionality reduction prior to classification. We observed consistently better performance using logistic regression and linear discriminant analysis models when compared to the other models we tested. Classification performance generally improved when an autoencoder was used, except for when we predicted the need for adjuvant treatment. We obtained overall best results using a logistic regression model without autoencoding to predict the need for adjuvant treatment (F1-score = 0.869).

Prediction of hospital readmission of multimorbid patients using machine learning models.

OBJECTIVE: The objective of this study is twofold. First, we seek to understand the characteristics of the multimorbid population that needs hospital care by using all diagnoses information (ICD-10 codes) and two aggregated multimorbidity and frailty scores. Second, we use machine learning prediction models on these multimorbid patients characteristics to predict rehospitalization within 30 and 365 days and their length of stay. METHODS: This study was conducted on 8 882 anonymized patients hospitalized at the University Hospital of Saint-Étienne. A descriptive statistical analysis was performed to better understand the characteristics of the patient population. Multimorbidity was measured using raw diagnoses information and two specific scores based on clusters of diagnoses: the Hospital Frailty Risk Score and the Calderon-Larrañaga index. Based on these variables different machine learning models (Decision Tree, Random forest and k-nearest Neighbors) were used to predict near future rehospitalization and length of stay (LoS). RESULTS: The use of random forest algorithms yielded better performance to predict both 365 and 30 days rehospitalization and using the diagnoses ICD-10 codes directly was significantly more efficient. However, using the Calderon-Larrañaga's clusters of diagnoses can be used as an efficient substitute for diagnoses information for predicting readmission. The predictive power of the algorithms is quite low on length of stay indicator. CONCLUSION: Using machine learning techniques using patients' diagnoses information and Calderon-Larrañaga's score yielded efficient results to predict hospital readmission of multimorbid patients. These methods could help improve the management of care of multimorbid patients in hospitals.

How We Can Reap the Full Benefit of Teleconsultations: Economic Evaluation Combined With a Performance Evaluation Through a Discrete-Event Simulation.

BACKGROUND: In recent years, the rapid development of information and communications technology enabled by innovations in videoconferencing solutions and the emergence of connected medical devices has contributed to expanding the scope of application and expediting the development of telemedicine. OBJECTIVE: This study evaluates the use of teleconsultations (TCs) for specialist consultations at hospitals in terms of costs, resource consumption, and patient travel time. The key feature of our evaluation framework is the combination of an economic evaluation through a cost analysis and a performance evaluation through a discrete-event simulation (DES) approach. METHODS: Three data sets were used to obtain detailed information on the characteristics of patients, characteristics of patients' residential locations, and usage of telehealth stations. A total of 532 patients who received at least one TC and 18,559 patients who received solely physical consultations (CSs) were included in the initial sample. The TC patients were recruited during a 7-month period (ie, 2020 data) versus 19 months for the CS patients (ie, 2019 and 2020 data). A propensity score matching procedure was applied in the economic evaluation. To identify the best scenarios for reaping the full benefits of TCs, various scenarios depicting different population types and deployment strategies were explored in the DES model. Associated break-even levels were calculated. RESULTS: The results of the cost evaluation reveal a higher cost for the TC group, mainly induced by higher volumes of (tele)consultations per patient and the substantial initial investment required for TC equipment. On average, the total cost per patient over 298 days of follow-up was €356.37 (US $392) per TC patient and €305.18 (US $336) per CS patient. However, the incremental cost of TCs was not statistically significant: €356.37 - €305.18 = €51.19 or US $392 - US $336 = US $56 (95% CI -35.99 to 114.25; P=.18). Sensitivity analysis suggested heterogeneous economic profitability levels within subpopulations and based on the intensity of use of TC solutions. In fact, the DES model results show that TCs could be a cost-saving strategy in some cases, depending on population characteristics, the amortization speed of telehealth equipment, and the locations of telehealth stations. CONCLUSIONS: The use of TCs has the potential to lead to a major organizational change in the health care system in the near future. Nevertheless, TC performance is strongly related to the context and deployment strategy involved.

How much does the hospital stay for infusion of anti-CD19 CAR-T cells cost to the French National Health Insurance?

Chimeric antigen receptor T-cells (CAR-T cells) have the potential to be a major innovation as a new type of cancer treatment, but are associated with extremely high prices and a high level of uncertainty. This study aims to assess the cost of the hospital stay for the administration of anti-CD19 CAR-T cells in France. Data were collected from the French Medical Information Systems Program (PMSI) and all hospital stays associated with an administrated drug encoded 9439938 (tisagenlecleucel, Kymriah®) or 9440456 (axicabtagene ciloleucel, Yescarta®) between January 2019 and December 2020 were included. 485 hospital stays associated with an injection of anti-CD19 CAR-T cells were identified, of which 44 (9%), 139 (28.7%), and 302 (62.3%) were for tisagenlecleucel in acute lymphoblastic leukaemia (ALL), tisagenlecleucel in diffuse large B-cell lymphoma (DLBCL), and axicabtagene ciloleucel respectively. The lengths of the stays were 37.9, 23.8, and 25.9 days for tisagenlecleucel in ALL, tisagenlecleucel in DLBCL, and axicabtagene ciloleucel, respectively. The mean costs per hospital stay were € 372,400 for a tisagenlecleucel in ALL, € 342,903 for tisagenlecleucel in DLBCL, and € 366,562 for axicabtagene ciloleucel. CAR T-cells represented more than 80% of these costs. n=13 hospitals performed CAR-T cell injections, with two hospitals accounting for more than 50% of the total number of injections. This study provides original data in a context of limited information regarding the costs of hospitalization for patients undergoing CAR-T cell treatments. In addition to the financial burden, distance may also be an important barrier for accessing CAR T-cell treatment.

Hospital Length of Stay Prediction Methods: A Systematic Review.

OBJECTIVE: This systematic review sought to establish a picture of length of stay (LOS) prediction methods based on available hospital data and study protocols designed to measure their performance. MATERIALS AND METHODS: An English literature search was done relative to hospital LOS prediction from 1972 to September 2019 according to the PRISMA guidelines. Articles were retrieved from PubMed, ScienceDirect, and arXiv databases. Information were extracted from the included papers according to a standardized assessment of population setting and study sample, data sources and input variables, LOS prediction methods, validation study design, and performance evaluation metrics. RESULTS: Among 74 selected articles, 98.6% (73/74) used patients' data to predict LOS; 27.0% (20/74) used temporal data; and 21.6% (16/74) used the data about hospitals. Overall, regressions were the most popular prediction methods (64.9%, 48/74), followed by machine learning (20.3%, 15/74) and deep learning (17.6%, 13/74). Regarding validation design, 35.1% (26/74) did not use a test set, whereas 47.3% (35/74) used a separate test set, and 17.6% (13/74) used cross-validation. The most used performance metrics were R2 (47.3%, 35/74), mean squared (or absolute) error (24.4%, 18/74), and the accuracy (14.9%, 11/74). Over the last decade, machine learning and deep learning methods became more popular (P=0.016), and test sets and cross-validation got more and more used (P=0.014). CONCLUSIONS: Methods to predict LOS are more and more elaborate and the assessment of their validity is increasingly rigorous. Reducing heterogeneity in how these methods are used and reported is key to transparency on their performance.

Incidence and risk factors for incisional hernia and recurrence: Retrospective analysis of the French national database.

AIM: The aim of this work was to determine the rate of incisional hernia (IH) repair and risk factors for IH repair after laparotomy. METHOD: This population-based study used data extracted from the French Programme de Médicalisation des Systèmes d'Informations (PMSI) database. All patients who had undergone a laparotomy in 2010, their hospital visits from 2010 to 2015 and patients who underwent a first IH repair in 2013 were included. Previously identified risk factors included age, gender, high blood pressure (HBP), obesity, diabetes and chronic obstructive pulmonary disease (COPD). RESULTS: Among the 431 619 patients who underwent a laparotomy in 2010, 5% underwent IH repair between 2010 and 2015. A high-risk list of the most frequent surgical procedures (>100) with a significant risk of IH repair (>10% at 5 years) was established and included 71 863 patients (17%; 65 procedures). The overall IH repair rate from this list was 17%. Gastrointestinal (GI) surgery represented 89% of procedures, with the majority of patients (72%) undergoing lower GI tract surgery. The IH repair rate was 56% at 1 year and 79% at 2 years. Risk factors for IH repair included obesity (31% vs 15% without obesity, p  < 0.001), COPD (20% vs 16% without COPD), HBP (19% vs 15% without HBP) and diabetes (19% vs 16% without diabetes). Obesity was the main risk factor for recurrence after IH repair (19% vs 13%, p < 0.001). CONCLUSION: From the PMSI database, the real rate of IH repair after laparotomy was 5%, increasing to 17% after digestive surgery. Obesity was the main risk factor, with an IH repair rate of 31% after digestive surgery. Because of the important medico-economic consequences, prevention of IH after laparotomy in high-risk patients should be considered.

Automatic and Explainable Labeling of Medical Event Logs With Autoencoding.

Process mining is a suitable method for knowledge extraction from patient pathways. Structured in event logs, medical events are complex, often described using various medical codes. An efficient labeling of these events before applying process mining analysis is challenging. This paper presents an innovative methodology to handle the complexity of events in medical event logs. Based on autoencoding, accurate labels are created by clustering similar events in latent space. Moreover, the explanation of created labels is provided by the decoding of its corresponding events. Tested on synthetic events, the method is able to find hidden clusters on sparse binary data, as well as accurately explain created labels. A case study on real healthcare data is performed. Results confirm the suitability of the method to extract knowledge from complex event logs representing patient pathways.

Comparison of two hospitalization modes in geriatrics either directly via a hotline or after an emergency unit admission.

Hospitalizations via the emergency services of the elderly represent on average 41% of the stays. The family physician is aware of the deleterious effects of using emergency rooms and know that intensive use contributes to the disorganization of these services. The provision of a telephone line, enabling doctors to have direct access to a geriatrician doctor, is a new service allowing, if necessary, direct hospitalization in geriatrics but its interest is still poorly evaluated. METHODS: From June 1st, 2015, to June 1st, 2016, we compared the route of care for inpatients directly in short stays of geriatrics thanks to the hotline (group hotline) versus the route of those passed by emergencies (group EU, emergency unit). RESULTS: 520 patients were included in the study. The duration of stay was shorter during hospitalization via the hotline, 11.6 [95% CI, 10.8-12.3] days in a direct hospitalization versus 14.1 [95% CI, 13.5-14.7] days of a passage through emergencies (p <0.05). Patients who were admitted to the emergency room were more likely to be hospitalized again. Among the 170 patients re-hospitalized, an average duration before re-hospitalization of 29.5 [CI 95%, 23.6-35.4] days was observed in patients hospitalised via the hotline, while those entered by emergencies were hospitalized in 24.1 [95% CI, 20.4-27.8] days (p <0.05). CONCLUSION: This analysis suggests that the intra-hospital course of geriatric patients directly addressed in short stays of geriatrics by direct admission was shorter and more efficient than the course of an intermediate stage in the emergencies. It seems important to discuss the generalization of the hotline device for the functioning of the geriatric pathway.

Performance and cost evaluation of health information systems using micro-costing and discrete-event simulation.

Innovation and health-care funding reforms have contributed to the deployment of Information and Communication Technology (ICT) to improve patient care. Many health-care organizations considered the application of ICT as a crucial key to enhance health-care management. The purpose of this paper is to provide a methodology to assess the organizational impact of high-level Health Information System (HIS) on patient pathway. We propose an integrated performance evaluation of HIS approach through the combination of formal modeling using the Architecture of Integrated Information Systems (ARIS) models, a micro-costing approach for cost evaluation, and a Discrete-Event Simulation (DES) approach. The methodology is applied to the consultation for cancer treatment process. Simulation scenarios are established to conclude about the impact of HIS on patient pathway. We demonstrated that although high level HIS lengthen the consultation, occupation rate of oncologists are lower and quality of service is higher (through the number of available information accessed during the consultation to formulate the diagnostic). The provided method allows also to determine the most cost-effective ICT elements to improve the care process quality while minimizing costs. The methodology is flexible enough to be applied to other health-care systems.

A stochastic optimization model for shift scheduling in emergency departments.

Excessive waiting time in Emergency Departments (ED) can be both a cause of frustration and more importantly, a health concern for patients. Waiting time arises when the demand for work goes beyond the facility's service capacity. ED service capacity mainly depends on human resources and on beds available for patients. In this paper, we focus on human resources organization in an ED and seek to best balance between service quality and working conditions. More specifically, we address the personnel scheduling problem in order to optimize the shift distribution among employees and minimize the total expected patients' waiting time. The problem is also characterized by a multi-stage re-entrant service process. With an appropriate approximation of patients' waiting times, we first propose a stochastic mixed-integer programming model that is solved by a sample average approximation (SAA) approach. The resulting personnel schedules are then evaluated using a discrete-event simulation model. Numerical experiments are then performed with data from a French hospital to compare different personnel scheduling strategies.

Modeling and simulation of blood collection systems.

This paper addresses the modeling and simulation of blood collection systems in France for both fixed site and mobile blood collection with walk in whole blood donors and scheduled plasma and platelet donors. Petri net models are first proposed to precisely describe different blood collection processes, donor behaviors, their material/human resource requirements and relevant regulations. Petri net models are then enriched with quantitative modeling of donor arrivals, donor behaviors, activity times and resource capacity. Relevant performance indicators are defined. The resulting simulation models can be straightforwardly implemented with any simulation language. Numerical experiments are performed to show how the simulation models can be used to select, for different walk in donor arrival patterns, appropriate human resource planning and donor appointment strategies.

Redesigning pharmacy delivery processes of a health care complex.

This paper addresses a pharmacy delivery design problem with two types of human resources: pharmacy assistants and transporters within a hospital. Each medical unit of the hospital has a mobile medicine closet which is conveyed each week by transporters to the central pharmacy for inventory assessment and refill by assistants. Transportation is carried out by foot, by tractor or by truck depending on the location. The problem consists in creating a transportation and supply planning for each day of the week in order to balance workloads for both transporters and assistants while ensuring the availability of medicine to each medical service. A two-step approach using mixed-integer linear programming formulation is proposed to determine a near optimal schedule. Numerical results are given to assess its efficiency. The proposed approach is then combined with a simulation model to redesign the delivery process of the pharmacy department of a French university teaching hospital. Methodology of this real-life reengineering study is presented and discussed.