RESUMO
The SARS-CoV-2 pandemic highlighted the need for software tools that could facilitate patient triage regarding potential disease severity or even death. In this article, an ensemble of Machine Learning (ML) algorithms is evaluated in terms of predicting the severity of their condition using plasma proteomics and clinical data as input. An overview of AI-based technical developments to support COVID-19 patient management is presented outlining the landscape of relevant technical developments. Based on this review, the use of an ensemble of ML algorithms that analyze clinical and biological data (i.e., plasma proteomics) of COVID-19 patients is designed and deployed to evaluate the potential use of AI for early COVID-19 patient triage. The proposed pipeline is evaluated using three publicly available datasets for training and testing. Three ML "tasks" are defined, and several algorithms are tested through a hyperparameter tuning method to identify the highest-performance models. As overfitting is one of the typical pitfalls for such approaches (mainly due to the size of the training/validation datasets), a variety of evaluation metrics are used to mitigate this risk. In the evaluation procedure, recall scores ranged from 0.6 to 0.74 and F1-score from 0.62 to 0.75. The best performance is observed via Multi-Layer Perceptron (MLP) and Support Vector Machines (SVM) algorithms. Additionally, input data (proteomics and clinical data) were ranked based on corresponding Shapley additive explanation (SHAP) values and evaluated for their prognosticated capacity and immuno-biological credence. This "interpretable" approach revealed that our ML models could discern critical COVID-19 cases predominantly based on patient's age and plasma proteins on B cell dysfunction, hyper-activation of inflammatory pathways like Toll-like receptors, and hypo-activation of developmental and immune pathways like SCF/c-Kit signaling. Finally, the herein computational workflow is corroborated in an independent dataset and MLP superiority along with the implication of the abovementioned predictive biological pathways are corroborated. Regarding limitations of the presented ML pipeline, the datasets used in this study contain less than 1000 observations and a significant number of input features hence constituting a high-dimensional low-sample (HDLS) dataset which could be sensitive to overfitting. An advantage of the proposed pipeline is that it combines biological data (plasma proteomics) with clinical-phenotypic data. Thus, in principle, the presented approach could enable patient triage in a timely fashion if used on already trained models. However, larger datasets and further systematic validation are needed to confirm the potential clinical value of this approach. The code is available on Github: https://github.com/inab-certh/Predicting-COVID-19-severity-through-interpretable-AI-analysis-of-plasma-proteomics.
Assuntos
Inteligência Artificial , COVID-19 , Humanos , COVID-19/diagnóstico , Aprendizado de Máquina , Proteômica , SARS-CoV-2RESUMO
BACKGROUND: Chronic respiratory conditions are a prominent public health issue and thus, building a patient registry might facilitate both policy decision making and improvement of clinical management processes. Hellenic Registry of patients with Home Mechanical Ventilation (HR-HMV) was initiated in 2017 and a web-based platform is used to support patient data collection. Eighteen hospital departments (including sleep labs) across Greece participate in this initiative, focusing on recording data for both children and adult patients supported by mechanical ventilation at home, including patients with Sleep Apnea-Hypopnea Syndrome (SAHS) under Positive Airway Pressure (PAP) therapy. METHODS: The HR-HMV initiative ultimately aims to provide a database for evidence-based care and policy making in this specific domain. To this end, a web information system was developed and data were manually collected by clinics and hospital departments. Legal and privacy issues (such as General Data Protection Rule compliance and technical information security measures) have been considered while designing the web application. Based on the collected data, an exploratory statistical report of SAHS patients in Greece is presented. RESULTS: Eleven out of the eighteen participating clinics and hospital departments have contributed with data by the time of the current study. More than 5000 adult and children patient records have been collected so far, the vast majority of which (i.e., 4900 patients) diagnosed with SAHS. CONCLUSION: The development and maintenance of patient registries is a valuable tool for policy decision making, observational/epidemiological research and beyond (e.g., health technology assessment procedures). However, as all data collection and processing approaches, registries are also related with potential biases. Along these lines, strengths and limitations must be considered when interpreting the collected data, and continuous validation of the collected clinical data per se should be emphasized. Especially for Greece, where the lack of national registries is eminent, we argue that HR-HMV could be a useful tool for the development and the update of related policies regarding the healthcare services for patients with home mechanical ventilation support and SAHS patients, which could be useful for related initiatives at a European level as well.
Assuntos
Serviços de Assistência Domiciliar , Apneia Obstrutiva do Sono , Adulto , Criança , Grécia , Humanos , Cooperação do Paciente , Sistema de Registros , Respiração ArtificialRESUMO
The value of social media data for Adverse Drug Reaction (ADR) monitoring is actively investigated. While social media provide a vast amount of data, these data are hard to analyse due to their unstructured nature and lack of credibility. Despite these challenges, social media have been identified as a potentially useful data source, potentially able to "strengthen" the evidence for new ADRs. To this end, PVClinical project aims to build a platform facilitating the investigation of multiple heterogeneous data sources, including social media, to support pharmacovigilance (PV) processes, both in the clinical environment and beyond. In this study, we present the PVClinical Twitter workspace, also highlighting the rationale behind the main design choices, while also discussing the respective challenges.
Assuntos
Efeitos Colaterais e Reações Adversas Relacionados a Medicamentos , Mídias Sociais , Sistemas de Notificação de Reações Adversas a Medicamentos , Humanos , Armazenamento e Recuperação da Informação , FarmacovigilânciaRESUMO
Partner Notification (PN) processes are typically part of wider combination prevention efforts and focus on the notification of sexual partners to prevent Sexually Transmitted Infections (STIs), including Human Immunodeficiency Viruses and viral hepatitis. We present a free, voluntary, anonymous and GDPR-compliant Partner Notification service that offers enhanced security and privacy through a web and mobile application via a unique random codes.
Assuntos
Infecções por HIV , Infecções Sexualmente Transmissíveis , Busca de Comunicante , Infecções por HIV/prevenção & controle , Humanos , Privacidade , Parceiros Sexuais , Infecções Sexualmente Transmissíveis/prevenção & controleRESUMO
Partner Notification processes focus on the notification of sexual partners to prevent the transmission of Sexually Transmitted Infections (STIs). The INTEGRATE Joint Action provides an integrated platform called RiskRadar, for combination prevention activities targeting STIs, including an anonymous, free and voluntary Partner Notification service. The presented service information flow ensures privacy, security and GDPR compliance which were identified as vital with similar tools. The service is available via web and mobile interfaces using a unique random code provided from authorised healthcare professionals to support privacy.
Assuntos
Busca de Comunicante , Infecções Sexualmente Transmissíveis , Segurança Computacional , Humanos , Parceiros Sexuais , Infecções Sexualmente Transmissíveis/prevenção & controleRESUMO
INTRODUCTION: Information technology (IT) plays an important role in the healthcare landscape via the increasing digitization of medical data and the use of modern computational paradigms such as machine learning (ML) and knowledge graphs (KGs). These 'intelligent' technical paradigms provide a new digital 'toolkit' supporting drug safety and healthcare processes, including 'active pharmacovigilance'. While these technical paradigms are promising, intelligent systems (ISs) are not yet widely adopted by pharmacovigilance (PV) stakeholders, namely the pharma industry, academia/research community, drug safety monitoring organizations, regulatory authorities, and healthcare institutions. The limitations obscuring the integration of ISs into PV activities are multifaceted, involving technical, legal and medical hurdles, and thus require further elucidation. OBJECTIVE: We dissect the abovementioned limitations by describing the lessons learned during the design and implementation of the PVClinical platform, a web platform aiming to support the investigation of potential adverse drug reactions (ADRs), emphasizing the use of knowledge engineering (KE) as its main technical paradigm. RESULTS: To this end, we elaborate on the related 'business processes' (i.e. operational processes) and 'user goals' identified as part of the PVClinical platform design process based on Design Thinking principles. We also elaborate on key challenges restricting the adoption of such ISs and their integration in the clinical setting and beyond. CONCLUSIONS: We highlight the fact that beyond providing analytics and useful statistics to the end user, 'actionability' has emerged as the operational priority identified through the whole process. Furthermore, we focus on the needs for valid, reproducible, explainable and human-interpretable results, stressing the need to emphasize on usability.
Assuntos
Efeitos Colaterais e Reações Adversas Relacionados a Medicamentos , Farmacovigilância , Atenção à Saúde , Humanos , Tecnologia da Informação , Aprendizado de MáquinaRESUMO
BACKGROUND: The HIV pandemic impacts the lives of millions and despite the global coordinated response, innovative actions are still needed to end it. A major challenge is the added burden of coinfections such as viral hepatitis, tuberculosis and various sexually transmitted infections in terms of prevention, treatment and increased morbidity in individuals with HIV infection. A need for combination prevention strategies, tailored to high-risk key populations arises and technology-based interventions can be a valuable asset. The COVID-19 pandemic challenged the delivery of existing services and added stress to existing public health and clinical structures but also highlighted the potential of exploiting technical solutions for interventions regarding infectious diseases. In this paper we report the design process, results and evaluation findings from the pilots of 'RiskRadar'-a web and mobile application aiming to support combination prevention, testing and linkage to care for HIV, viral hepatitis, various sexually transmitted infections and tuberculosis. METHODS: RiskRadar was developed for the INTEGRATE Joint Action's aim to improve, adapt and pilot innovative digital tools for combination prevention. RiskRadar was designed iteratively using informed end-user-oriented approaches. Emphasis was placed on the Risk Calculator that enables users to assess their risk of exposure to one or more of the four disease areas, make informed decisions to seek testing or care and adjust their behaviours ultimately aiming to harm/risk reduction. RiskRadar has been piloted in three countries, namely Croatia, Italy and Lithuania. RESULTS: RiskRadar has been used 1347 times across all platforms so far. More than 90% of users have found RiskRadar useful and would use it again, especially the Risk Calculator component. Almost 49.25% are men and 29.85% are in the age group of 25-34. The application has scored 5.2/7 in the User Experience Questionnaire, where it is mainly described as "supportive" and "easy-to-use". The qualitative evaluation of RiskRadar also yielded positive feedback. CONCLUSIONS: Pilot results demonstrate above average satisfaction with RiskRadar and high user-reported usability scores, supporting the idea that technical interventions could significantly support combination prevention actions on Sexually Transmitted Infections.
Assuntos
COVID-19 , Infecções por HIV , Hepatite Viral Humana , Infecções Sexualmente Transmissíveis , Tuberculose , Adulto , Infecções por HIV/epidemiologia , Infecções por HIV/prevenção & controle , Hepatite Viral Humana/epidemiologia , Hepatite Viral Humana/prevenção & controle , Humanos , Masculino , Pandemias , SARS-CoV-2 , Infecções Sexualmente Transmissíveis/epidemiologia , Infecções Sexualmente Transmissíveis/prevenção & controle , Tuberculose/prevenção & controleRESUMO
Information Technology (IT) and specialized systems could have a prominent role towards the support of drug safety processes, both in the clinical context but also beyond that. PVClinical project aims to build an IT platform, enabling the investigation of potential Adverse Drug Reactions (ADRs). In this paper, we outline the utilization of Observational Medical Outcomes Partnership - Common Data Model (OMOP-CDM) and the openly available Observational Health Data Sciences and Informatics (OHDSI) software stack as part of PVClinical platform. OMOP-CDM offers the capacity to integrate data from Electronic Health Records (EHRs) (e.g., encounters, patients, providers, diagnoses, drugs, measurements and procedures) via an accepted data model. Furthermore, the OHDSI software stack provides valuable analytics tools which could be used to address important questions regarding drug safety quickly and efficiently, enabling the investigation of potential ADRs in the clinical environment.
Assuntos
Informática Médica , Farmacovigilância , Ciência de Dados , Bases de Dados Factuais , Registros Eletrônicos de Saúde , Humanos , SoftwareRESUMO
Clinical Decision Support Systems (CDSS) could play a prominent role in preventing Adverse Drug Reactions (ADRs) especially when integrated in larger healthcare systems (e.g. Electronic Health Record - EHR systems, Hospital Management Systems - HMS, e-Prescription systems etc.). This poster presents an approach to model Therapeutic Prescription Protocols (TPPs) via the Business Process Management Notation (BPMN), as part of the e-Prescription CDSS developed in the context of the PrescIT project.
Assuntos
Sistemas de Apoio a Decisões Clínicas , Efeitos Colaterais e Reações Adversas Relacionados a Medicamentos , Sistemas Computacionais , Atenção à Saúde , Humanos , PrescriçõesRESUMO
Online digital tools are considered an innovative method to promote HIV, hepatitis and STIs prevention, testing and treatment services, overcoming individual and social barriers, especially for younger people and other, possibly hard-to-reach, target population groups. In this paper, we introduce INTEGRATE RiskRadar, a web and mobile application developed in the scope of the EU-supported INTEGRATE Joint Action (JA), that aims to enhance the integration of combination prevention, testing and linkage to care for HIV, hepatitis, STIs and tuberculosis by providing integrated information and digital tools regarding all four diseases to population groups at increased risk, aiming to eliminate the individual and social barriers to effective adoption of prevention practices, testing and linkage to care, and thus reduce the incidence and burden of these diseases in the European Region.