Evaluating distributed-learning on real-world obstetrics data: comparing distributed, centralized and local models.

This study focused on comparing distributed learning models with centralized and local models, assessing their efficacy in predicting specific delivery and patient-related outcomes in obstetrics using real-world data. The predictions focus on key moments in the obstetric care process, including discharge and various stages of hospitalization. Our analysis: using 6 different machine learning methods like Decision Trees, Bayesian methods, Stochastic Gradient Descent, K-nearest neighbors, AdaBoost, and Multi-layer Perceptron and 19 different variables with various distributions and types, revealed that distributed models were at least equal, and often superior, to centralized versions and local versions. We also describe thoroughly the preprocessing stage in order to help others implement this method in real-world scenarios. The preprocessing steps included cleaning and harmonizing missing values, handling missing data and encoding categorical variables with multisite logic. Even though the type of machine learning model and the distribution of the outcome variable can impact the result, we reached results of 66% being superior to the centralized and local counterpart and 77% being better than the centralized with AdaBoost. Our experiments also shed light in the preprocessing steps required to implement distributed models in a real-world scenario. Our results advocate for distributed learning as a promising tool for applying machine learning in clinical settings, particularly when privacy and data security are paramount, thus offering a robust solution for privacy-concerned clinical applications.

FHIR Based Interoperability of Medical Devices.

Although FHIR has been designed to be easy to implement, it requires knowledge that is still hard to find. We aim to evaluate the use of FHIR in Portuguese projects for the integration of medical devices. Two projects were selected, including easyHealth4Covid (EH4C) and Chronic Diseases Management Platform (CDMP). The evolution of each project and the FHIR resources used were analyzed. 11 different sensors of 5 companies were used in the sum of both projects. Previously, none of them used FHIR to integrate and the teams had little to no experience in doing so. The FHIR Observation resource was used for all. There is a general lack of knowledge of the FHIR standard and terminologies of most of the device companies involved in the projects.

Dataset Comparison Tool: Utility and Privacy.

Synthetic data has been more and more used in the last few years. While its applications are various, measuring its utility and privacy is seldom an easy task. Since there are different methods of evaluating these issues, which are dependent on data types, use cases and purpose, a generic method for evaluating utility and privacy does not exist at the moment. So, we introduced a compilation of the most recent methods for evaluating privacy and utility into a single executable in order to create a report of the similarities and potential privacy breaches between two datasets, whether it is related to synthetic or not. We catalogued 24 different methods, from qualitative to quantitative, column-wise or table-wise evaluations. We hope this resource can help scientists and industries get a better grasp of the synthetic data they have and produce more easily and a better basis to create a new, more broad method for evaluating dataset similarities.

Pregnant Users' Perceptions of the Birth Plan Interface in the "My Prenatal Care" App: Observational Validation Study.

BACKGROUND: Birth plans are meant to be a declaration of the expectations and preferences of pregnant woman regarding childbirth. The My Prenatal Care app engages pregnant women in an educational intervention for a healthy pregnancy. We hypothesized that users' positive perception of an in-app birth plan is a relevant step for establishing direct communication between pregnant women and the health care team, based on an online report available on the app. OBJECTIVE: This study aimed to evaluate pregnant women's perception about the communicability of birth-plan preparation using a mobile app. METHODS: This was an observational, exploratory, descriptive study. The methodology was user centered, and both qualitative and quantitative approaches were employed. The tools of the communicability evaluation method were applied. Overall, 11 pregnant women evaluated their experience of using a birth-plan prototype interface. The evaluation was performed in a controlled environment, with authorized video recording. There were 8 task-oriented interactions proposed to evaluate interface communicability with users when using the Birth Plan menu. For evaluating perceptions and experiences, a survey with structured and open-ended questions in addition to the free expression of participants was conducted. The primary outcomes assessed were interface communicability and user's perception of the Birth Plan prototype interface in the My Prenatal Care mobile app. Secondarily, we involved users in the prototyping phase of the interface to identify bottlenecks for making improvements in the app. RESULTS: Regarding users' performance in accomplishing previously prepared tasks, we found that 10 of 11 (91%) women were capable of completing at least 6 of 8 (75%) tasks. A positive relationship was found between the number of communicability problems and the success of completing the tasks. An analysis of the records revealed three communicability breakdowns related to the data entry, save, and scrollbar functions. The participants freely expressed suggestions for improvements such as for the save function and the process of sharing the birth-plan form upon completion. CONCLUSIONS: Users had a positive perception of the Birth Plan menu of the My Prenatal Care app. This user-centered validation enabled the identification of solutions for problems, resulting in improvements in the app.

Nutrigenomic Information in the openEHR Data Set.

BACKGROUND: The traditional concept of personalized nutrition is based on adapting diets according to individual needs and preferences. Discussions about personalized nutrition have been on since the Human Genome Project, which has sequenced the human genome. Thenceforth, topics such as nutrigenomics have been assessed to help in better understanding the genetic variation influence on the dietary response and association between nutrients and gene expression. Hence, some challenges impaired the understanding about the nowadays important clinical data and about clinical data assumed to be important in the future. OBJECTIVE: Finding the main clinical statements in the personalized nutrition field (nutrigenomics) to create the future-proof health information system to the openEHR server based on archetypes, as well as a specific nutrigenomic template. METHODS: A systematic literature search was conducted in electronic databases such as PubMed. The aim of this systemic review was to list the chief clinical statements and create archetype and templates for openEHR modeling tools, namely, Ocean Archetype Editor and Ocean Template Design. RESULTS: The literature search led to 51 articles; however, just 26 articles were analyzed after all the herein adopted inclusion criteria were assessed. Of these total, 117 clinical statements were identified, as well as 27 archetype-friendly concepts. Our group modeled four new archetypes (waist-to-height ratio, genetic test results, genetic summary, and diet plan) and finally created the specific nutrigenomic template for nutrition care. CONCLUSION: The archetypes and the specific openEHR template developed in this study gave dieticians and other health professionals an important tool to their nutrigenomic clinical practices, besides a set of nutrigenomic data to clinical research.

Identification and Characterization of Inter-Organizational Information Flows in the Portuguese National Health Service.

OBJECTIVES: To understand and build a collective vision of all existing institutions in the Portuguese National Health Service as well as to perceive how and how far the interaction between those multiple institutions is supported by Information Systems (IS). METHODS: Upon identification of the institutions involved in the healthcare process, a set of interviews with experienced people from those institutions was conducted, which produced about five hours of tape. The research was focused exclusively on processes involving two different organizations and any internal processes were altogether excluded from it. RESULTS: The study allowed the identification of about 50 recurrent interaction processes, which were classified into four different varieties in accordance with the nature of the information flow: administrative, clinical, identificational and statistical. In addition, these processes were divided in accordance with the way how that integration is achieved, from completely automated to email or telephone-based. CONCLUSIONS: Funds/Money related processes are technologically more rigid and standardized, whereas auditing and inspection ones are less supported by automatic systems. There emerged an interesting level of sharing and integration in clinical processes, although the integration is mostly made at the interface level. The authors identified 5 particularly relevant and dominant actors (2 classes of individuals and 3 institutions) with which there is a need for coordination and cooperation. The authors consider that, in future works, an effort should be made to provide the various institutions with guidelines/interfaces and prompt such institutions to elaborate upon these.

[Validity of classical fetal weight charts in the Portuguese population]. / Validade das tabelas de peso fetal clássicas para a população Portuguesa.

PURPOSE: to assess the validity of several fetal weight charts, commonly used in Portugal, to classify its population. METHODS: observational retrospective study. Singleton birth data was analyzed, from a two- year period (May 2008 to April 2010), from pregnancies with an ultrasound in the same institution, between the 8th and 14th gestational week. Upon data validation, percentiles for each completed gestational week were created, smoothed by a quadratic function, analyzed and compared to the tables more commonly utilized, in the institution and country, by using Z-scores, percentile comparison, sample 10th percentile detection sensibility and birthweight means comparison. RESULTS: a total of 5,378 newborns (NB) were born in the period; 2,195 (42%) NB were included, born from the 24th to 42nd gestational week, allowing statistical analysis from the 34th to the 41st week. There were differences in the mean birthweight for each gestational age, between references and with the sample, as well as between sexes. The 10th percentile from some references has shown differences ranging from -288g at 37 weeks (-11% in Lubchenco et al. data), with and +133g at 34 weeks (+7,6% with Carrascosa et al. data) compared to the values found with the sample. Differences were also found concerning the sensitivity of the identification of a sample birthweight below the 10th percentile, which was between 14.1 and 100%, depending on the reference used. DISCUSSION: the limitation of these kinds of reference values must be remembered and minimized, with the adoption of regionally or nationally produced references, contemplating other variables, such as sex, with precisely known gestation duration and with validation of the utilized references in loco.

Validade das tabelas de peso fetal clássicas para a população Portuguesa / Validity of classical fetal weight charts in the Portuguese population

OBJETIVO: avaliar a validade de várias tabelas de peso fetal, habitualmente usadas em Portugal, para classificar a sua população. MÉTODOS: estudo observacional retrospectivo. Foram analisados os registos de nascimentos no período de dois anos (Maio de 2008 a Abril de 2010), decorrentes de gestações unifetais com datação precisa por ecografia entre as 8ª e 14ª semanas de gestação, na mesma instituição. Após validação dos registos, foram analisados os percentis de peso gerados para cada semana de gestação completa, suavizados por uma função polinomial de dois graus, comparando-os com as tabelas mais usadas na instituição e no país, através do uso de Z-scores, valores de percentis, sensibilidade para detecção do percentil 10 (P10) da amostra e comparação de médias de peso. RESULTADOS: um total de 5.378 recém-nascidos (RN) foi registado no período; 2.195 (42 por cento) RN corresponderam aos critérios de inclusão, com idade gestacional (IG) entre as 24ª e 42ª semanas, permitindo uma análise estatística entre as 34ª e as 41ª semanas. Foram detectadas diferenças no peso médio por IG entre tabelas e em relação à amostra, bem como diferenças entre os sexos. O P10 de outros trabalhos mostrou diferenças entre -288g na 37ªs (-11 por cento nos dados de Lubchenco et al.) e +133g na 34ªs (+7,6 por cento nos dados de Carrascosa et al.) em relação ao obtido na amostra. A sensibilidade para detecção de um RN abaixo do P10 na amostra variou, às 39ªs, entre 14,1e 100 por cento, dependendo da tabela usada. DISCUSSÃO: as limitações deste tipo de valores de referência devem ser tidas em consideração, tentando minimizá-las, nomeadamente pela criação de valores locais/regionais ou nacionais, com a contemplação de outras variáveis, sobretudo o sexo do RN, em gravidezes rigorosamente datadas e pela validação in loco dos valores utilizados.

PURPOSE: to assess the validity of several fetal weight charts, commonly used in Portugal, to classify its population. METHODS: observational retrospective study. Singleton birth data was analyzed, from a two- year period (May 2008 to April 2010), from pregnancies with an ultrasound in the same institution, between the 8th and 14th gestational week. Upon data validation, percentiles for each completed gestational week were created, smoothed by a quadratic function, analyzed and compared to the tables more commonly utilized, in the institution and country, by using Z-scores, percentile comparison, sample 10th percentile detection sensibility and birthweight means comparison. RESULTS: a total of 5,378 newborns (NB) were born in the period; 2,195 (42 percent) NB were included, born from the 24th to 42nd gestational week, allowing statistical analysis from the 34th to the 41st week. There were differences in the mean birthweight for each gestational age, between references and with the sample, as well as between sexes. The 10th percentile from some references has shown differences ranging from -288g at 37 weeks (-11 percent in Lubchenco et al. data), with and +133g at 34 weeks (+7,6 percent with Carrascosa et al. data) compared to the values found with the sample. Differences were also found concerning the sensitivity of the identification of a sample birthweight below the 10th percentile, which was between 14.1 and 100 percent, depending on the reference used. DISCUSSION: the limitation of these kinds of reference values must be remembered and minimized, with the adoption of regionally or nationally produced references, contemplating other variables, such as sex, with precisely known gestation duration and with validation of the utilized references in loco.

Implementation, monitoring and utilization of an integrated Hospital Information System--lessons from a case study.

In most hospitals several heterogeneous Information Systems (IS) store parts of a still scattered patient record. Virtual Patient Records (VPR) are systems that aggregate known data elements about the patient from different IS in real-time. This papers aims to present the main lessons learned from the implementation and the usage during 6 years of a VPR system. Ten major lessons were divided in recommendations for software developers, information managers and institutional policy makers. Implementing and using a VPR is a difficult journey but can generate great value for the institution if most of these recommendations are taken in consideration.