Comparing Decentralized Learning Methods for Health Data Models to Nondecentralized Alternatives: Protocol for a Systematic Review.

BACKGROUND: Considering the soaring health-related costs directed toward a growing, aging, and comorbid population, the health sector needs effective data-driven interventions while managing rising care costs. While health interventions using data mining have become more robust and adopted, they often demand high-quality big data. However, growing privacy concerns have hindered large-scale data sharing. In parallel, recently introduced legal instruments require complex implementations, especially when it comes to biomedical data. New privacy-preserving technologies, such as decentralized learning, make it possible to create health models without mobilizing data sets by using distributed computation principles. Several multinational partnerships, including a recent agreement between the United States and the European Union, are adopting these techniques for next-generation data science. While these approaches are promising, there is no clear and robust evidence synthesis of health care applications. OBJECTIVE: The main aim is to compare the performance among health data models (eg, automated diagnosis and mortality prediction) developed using decentralized learning approaches (eg, federated and blockchain) to those using centralized or local methods. Secondary aims are comparing the privacy compromise and resource use among model architectures. METHODS: We will conduct a systematic review using the first-ever registered research protocol for this topic following a robust search methodology, including several biomedical and computational databases. This work will compare health data models differing in development architecture, grouping them according to their clinical applications. For reporting purposes, a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 flow diagram will be presented. CHARMS (Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies)-based forms will be used for data extraction and to assess the risk of bias, alongside PROBAST (Prediction Model Risk of Bias Assessment Tool). All effect measures in the original studies will be reported. RESULTS: The queries and data extractions are expected to start on February 28, 2023, and end by July 31, 2023. The research protocol was registered with PROSPERO, under the number 393126, on February 3, 2023. With this protocol, we detail how we will conduct the systematic review. With that study, we aim to summarize the progress and findings from state-of-the-art decentralized learning models in health care in comparison to their local and centralized counterparts. Results are expected to clarify the consensuses and heterogeneities reported and help guide the research and development of new robust and sustainable applications to address the health data privacy problem, with applicability in real-world settings. CONCLUSIONS: We expect to clearly present the status quo of these privacy-preserving technologies in health care. With this robust synthesis of the currently available scientific evidence, the review will inform health technology assessment and evidence-based decisions, from health professionals, data scientists, and policy makers alike. Importantly, it should also guide the development and application of new tools in service of patients' privacy and future research. TRIAL REGISTRATION: PROSPERO 393126; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=393126. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): PRR1-10.2196/45823.

Secur-e-Health Project: Towards Federated Learning for Smart Pediatric Care.

The application of machine learning (ML) algorithms to electronic health records (EHR) data allows the achievement of data-driven insights on various clinical problems and the development of clinical decision support (CDS) systems to improve patient care. However, data governance and privacy barriers hinder the use of data from multiple sources, especially in the medical field due to the sensitivity of data. Federated learning (FL) is an attractive data privacy-preserving solution in this context by enabling the training of ML models with data from multiple sources without any data sharing, using distributed remotely hosted datasets. The Secur-e-Health project aims at developing a solution in terms of CDS tools encompassing FL predictive models and recommendation systems. This tool may be especially useful in Pediatrics due to the increasing demands on Pediatric services, and the current scarcity of ML applications in this field compared to adult care. Herein we provide a description of the technical solution proposed in this project for three specific pediatric clinical problems: childhood obesity management, pilonidal cyst post-surgical care and retinography imaging analysis.

Low Memory T Cells Blood Counts and High Naïve Regulatory T Cells Percentage at Relapsing Remitting Multiple Sclerosis Diagnosis.

Objective: The aim of this study is to assess the peripheral immune system of newly diagnosed patients with relapsing remitting multiple sclerosis (RRMS) and compare it to healthy controls (HC). Methods: This cross-sectional study involves 30 treatment-naïve newly diagnosed patients with RRMS and 33 sex- and age-matched HC. Peripheral blood mononuclear cells were analyzed regarding: i) thymic function surrogates [T cell receptor excision circles (TRECs) and recent thymic emigrants (RTEs)]; ii) naïve and memory CD4+ and CD8+ T cells subsets; iii) T helper (Th) phenotype and chemokine receptors expression on CD8+ T cells subsets; iv) regulatory T cell (Tregs) phenotype; and exclude expression of activating/inhibitory receptors by natural killer (NK) and NKT cells. Analyses were controlled for age, sex, and human cytomegalovirus (HCMV) IgG seroprevalence. Results: Newly diagnosed patients with RRMS and HC have equivalent thymic function as determined by similar numbers of RTEs and levels of sjTRECs, DJßTRECs, and sj/DJßTREC ratio. In the CD8+ T cells compartment, patients with RRMS have a higher naive to memory ratio and lower memory cell counts in blood, specifically of effector memory and TemRA CD8+ T cells. Interestingly, higher numbers and percentages of central memory CD8+ T cells are associated with increasing time from the relapse. Among CD4+ T cells, lower blood counts of effector memory cells are found in patients upon controlling for sex, age, and anti-HCMV IgG seroprevalence. Higher numbers of CD4+ T cells (both naïve and memory) and of Th2 cells are associated with increasing time from the relapse; lower numbers of Th17 cells are associated with higher MS severity scores (MSSS). Patients with RRMS have a higher percentage of naïve Tregs compared with HC, and lower percentages of these cells are associated with higher MSSS. Percentages of immature CD56bright NK cells expressing the inhibitory receptor KLRG1 and of mature CD56dimCD57+ NK cells expressing NKp30 are higher in patients. No major alterations are observed on NKT cells. Conclusion: Characterization of the peripheral immune system of treatment-naïve newly diagnosed patients with RRMS unveiled immune features present at clinical onset including lower memory T cells blood counts, particularly among CD8+ T cells, higher percentage of naïve Tregs and altered percentages of NK cells subsets expressing inhibitory or activating receptors. These findings might set the basis to better understand disease pathogenesis.

Case Report: Cryptococcal meningitis in Hodgkin's Lymphoma patient receiving brentuximab-vedotin therapy.

Cryptococcus neoformans infections occur mostly in immunodeficient individuals, being the most common opportunistic fungi central nervous system (CNS) infection in HIV seropositive patients. Moreover, other conditions affecting host immunity, such as hematologic malignancies, organ transplantation and immunosuppressive drugs are implicated as risk factors. The authors present a case of a 48-year-old male with Hodgkin Lymphoma for 26 years and submitted to several lines of treatment, diagnosed with cryptococcal meningitis while on therapy with brentuximab. The patient presented with positive cerebral spinal fluid (CSF) cryptococcal antigen plus positive blood cultures. He was put under induction antifungal treatment with liposomal amphotericin B and flucytosine, as well as corticotherapy with dexamethasone with headache improvement and a favorable clinical evolution. There are no reported cases of cryptococcal meningoencephalitis under CD30-directed monoclonal antibody. Furthermore, this case illustrates the risk of Cryptococcus neoformans infection in immunocompromising conditions other than HIV, underlining the need of considering this differential diagnosis when physicians face an opportunist neuroinfection.

Definition of Immunological Nonresponse to Antiretroviral Therapy: A Systematic Review.

BACKGROUND: Terms and criteria to classify people living with HIV on antiretroviral therapy who fail to achieve satisfactory CD4 T-cell counts are heterogeneous, and need revision and summarization. METHODS: We performed a systematic review of PubMed original research articles containing a set of predefined terms, published in English between January 2009 and September 2018. The search retrieved initially 1360 studies, of which 103 were eligible. The representative terminology and criteria were extracted and analyzed. RESULTS: Twenty-two terms and 73 criteria to define the condition were identified. The most frequent term was "immunological nonresponders" and the most frequent criterion was "CD4 T-cell count <350 cells/µL after ≥24 months of virologic suppression." Most criteria use CD4+ T-cell counts as a surrogate, either as an absolute value or as a change after a defined period of time [corrected]. Distinct values and time points were used. Only 9 of the 73 criteria were used by more than one independent research team. Herein we propose 2 criteria that could help to reach a consensus. CONCLUSIONS: The high disparity in terms and criteria here reported precludes data aggregation and progression of the knowledge on this condition, because it renders impossible to compare data from different studies. This review will foster the discussion of terms and criteria to achieve a consensual definition.

Thymic Function as a Predictor of Immune Recovery in Chronically HIV-Infected Patients Initiating Antiretroviral Therapy.

Poor immunological responders (PIR) are HIV-infected patients with virologic suppression upon antiretroviral therapy (ART) but persistently low CD4+ T cell counts. Early identification of PIR is important given their higher morbimortality compared to adequate immune responders (AIR). In this study, 33 patients severely lymphopenic at ART onset, were followed for at least 36 months, and classified as PIR or AIR using cluster analysis grounded on their CD4+ T cell count trajectories. Based on a variety of immunological parameters, we built predictive models of PIR/AIR outcome using logistic regression. All PIR had CD4+ T cell counts consistently below 500 cells/µL, while all AIR reached this threshold. AIR showed a higher percentage of recent thymic emigrants among CD4+ T cells; higher numbers of sj-TRECs and greater sj/ß TREC ratios; and significant increases in thymic volume from baseline to 12 months of ART. We identified mathematical models that correctly predicted PIR/AIR outcome after 36 months of therapy in 77-87% of the cases, based on observations made until 2-6 months after ART onset. This study highlights the importance of thymic activity in the immune recovery of severely lymphopenic patients, and may help to select the patients that will benefit from closer follow-up or novel therapeutic approaches.

Toxoplasmosis-associated IRIS involving the CNS: a case report with longitudinal analysis of T cell subsets.

BACKGROUND: HIV-infected patients may present an unforeseen clinical worsening after initiating antiretroviral therapy known as immune reconstitution inflammatory syndrome (IRIS). This syndrome is characterized by a heightened inflammatory response toward infectious or non-infectious triggers, and it may affect different organs. Diagnosis of IRIS involving the central nervous system (CNS-IRIS) is challenging due to heterogeneous manifestations, absence of biomarkers to identify this condition, risk of long-term sequelae and high mortality. Hence, a deeper knowledge of CNS-IRIS pathogenesis is needed. CASE PRESENTATION: A 37-year-old man was diagnosed with AIDS and cerebral toxoplasmosis. Anti-toxoplasma treatment was initiated immediately, followed by active antiretroviral therapy (HAART) 1 month later. At 2 months of HAART, he presented with progressive hyposensitivity of the right lower limb associated with brain and dorsal spinal cord lesions, compatible with paradoxical toxoplasmosis-associated CNS-IRIS, a condition with very few reported cases. A stereotactic biopsy was planned but was postponed based on its inherent risks. Patient showed clinical improvement with no requirement of corticosteroid therapy. Routine laboratorial analysis was complemented with longitudinal evaluation of blood T cell subsets at 0, 1, 2, 3 and 6 months upon HAART initiation. A control group composed by 9 HIV-infected patients from the same hospital but with no IRIS was analysed for comparison. The CNS-IRIS patient showed lower percentage of memory CD4+ T cells and higher percentage of activated CD4+ T cells at HAART initiation. The percentage of memory CD4+ T cells drastically increased at 1 month after HAART initiation and became higher in comparison to the control group until clinical recovery onset; the percentage of memory CD8+ T cells was consistently lower throughout follow-up. Interestingly, the percentage of regulatory T cells (Treg) on the CNS-IRIS patient reached a minimum around 1 month before symptoms onset. CONCLUSION: Although both stereotactic biopsies and steroid therapy might be of use in CNS-IRIS cases and should be considered for these patients, they might be unnecessary to achieve clinical improvement as shown in this case. Immunological characterization of more CNS-IRIS cases is essential to shed some light on the pathogenesis of this condition.

Longitudinal evaluation of regulatory T-cell dynamics on HIV-infected individuals during the first 2 years of therapy.

OBJECTIVES: A sizeable percentage of individuals infected by HIV and on antiretroviral therapy (ART) fail to increase their CD4 T-cells to satisfactory levels. The percentage of regulatory T-cells (Tregs) has been suggested to contribute to this impairment. This study aimed to address this question and to expand the analysis of Tregs subpopulations during ART. DESIGN: Longitudinal follow-up of 81 HIV-infected individuals during the first 24 months on ART. METHODS: CD4 T-cell counts, Tregs percentages, and specific Tregs subpopulations were evaluated at ART onset, 2, 6, 9, 12, 16, 20, and 24 months of ART (five individuals had no Tregs information at baseline). RESULTS: The slope of CD4 T-cell recovery was similar for individuals with moderate and with severe lymphopenia at ART onset. No evidence was found for a contribution of the baseline Tregs percentages on the CD4 T-cell counts recovery throughout ART. In comparison to uninfected individuals, Tregs percentages were higher at ART onset only for patients with less than 200 cells/µl at baseline and decreased afterwards reaching normal values. Within Tregs, the percentage of naive cells remained low in these patients. Reduced thymic export and increased proliferation of Tregs vs. conventional CD4 T cells might explain these persistent alterations. CONCLUSION: No effect of Tregs percentages at baseline was detected on CD4 T-cell recovery. However, profound alterations on Tregs subpopulations were consistently observed throughout ART for patients with severe lymphopenia at ART onset.