A multi-ethnic reference panel to impute HLA classical and non-classical class I alleles in admixed samples: Testing imputation accuracy in an admixed sample from Brazil.

The MHC class I region contains crucial genes for the innate and adaptive immune response, playing a key role in susceptibility to many autoimmune and infectious diseases. Genome-wide association studies have identified numerous disease-associated SNPs within this region. However, these associations do not fully capture the immune-biological relevance of specific HLA alleles. HLA imputation techniques may leverage available SNP arrays by predicting allele genotypes based on the linkage disequilibrium between SNPs and specific HLA alleles. Successful imputation requires diverse and large reference panels, especially for admixed populations. This study employed a bioinformatics approach to call SNPs and HLA alleles in multi-ethnic samples from the 1000 genomes (1KG) dataset and admixed individuals from Brazil (SABE), utilising 30X whole-genome sequencing data. Using HIBAG, we created three reference panels: 1KG (n = 2504), SABE (n = 1171), and the full model (n = 3675) encompassing all samples. In extensive cross-validation of these reference panels, the multi-ethnic 1KG reference exhibited overall superior performance than the reference with only Brazilian samples. However, the best results were achieved with the full model. Additionally, we expanded the scope of imputation by developing reference panels for non-classical, MICA, MICB and HLA-H genes, previously unavailable for multi-ethnic populations. Validation in an independent Brazilian dataset showcased the superiority of our reference panels over the Michigan Imputation Server, particularly in predicting HLA-B alleles among Brazilians. Our investigations underscored the need to enhance or adapt reference panels to encompass the target population's genetic diversity, emphasising the significance of multiethnic references for accurate imputation across different populations.

Design, cohort profile and comparison of the KTD-Innov study: a prospective multidimensional biomarker validation study in kidney allograft rejection.

There is an unmet need for robust and clinically validated biomarkers of kidney allograft rejection. Here we present the KTD-Innov study (ClinicalTrials.gov, NCT03582436), an unselected deeply phenotyped cohort of kidney transplant recipients with a holistic approach to validate the clinical utility of precision diagnostic biomarkers. In 2018-2019, we prospectively enrolled consecutive adult patients who received a kidney allograft at seven French centers and followed them for a year. We performed multimodal phenotyping at follow-up visits, by collecting clinical, biological, immunological, and histological parameters, and analyzing a panel of 147 blood, urinary and kidney tissue biomarkers. The primary outcome was allograft rejection, assessed at each visit according to the international Banff 2019 classification. We evaluated the representativeness of participants by comparing them with patients from French, European, and American transplant programs transplanted during the same period. A total of 733 kidney transplant recipients (64.1% male and 35.9% female) were included during the study. The median follow-up after transplantation was 12.3 months (interquartile range, 11.9-13.1 months). The cumulative incidence of rejection was 9.7% at one year post-transplant. We developed a distributed and secured data repository in compliance with the general data protection regulation. We established a multimodal biomarker biobank of 16,736 samples, including 9331 blood, 4425 urinary and 2980 kidney tissue samples, managed and secured in a collaborative network involving 7 clinical centers, 4 analytical platforms and 2 industrial partners. Patients' characteristics, immune profiles and treatments closely resembled those of 41,238 French, European and American kidney transplant recipients. The KTD-Innov study is a unique holistic and multidimensional biomarker validation cohort of kidney transplant recipients representative of the real-world transplant population. Future findings from this cohort are likely to be robust and generalizable.

Sepsis-trained macrophages promote antitumoral tissue-resident T cells.

Sepsis induces immune alterations, which last for months after the resolution of illness. The effect of this immunological reprogramming on the risk of developing cancer is unclear. Here we use a national claims database to show that sepsis survivors had a lower cumulative incidence of cancers than matched nonsevere infection survivors. We identify a chemokine network released from sepsis-trained resident macrophages that triggers tissue residency of T cells via CCR2 and CXCR6 stimulations as the immune mechanism responsible for this decreased risk of de novo tumor development after sepsis cure. While nonseptic inflammation does not provoke this network, laminarin injection could therapeutically reproduce the protective sepsis effect. This chemokine network and CXCR6 tissue-resident T cell accumulation were detected in humans with sepsis and were associated with prolonged survival in humans with cancer. These findings identify a therapeutically relevant antitumor consequence of sepsis-induced trained immunity.

Mortality in a Nationwide Practice-Based Cohort Receiving Paclitaxel-Coated Devices for Lower Limb Peripheral Artery Disease.

BACKGROUND: According to a meta-analysis of randomized clinical trials, paclitaxel-coated devices (PCDs) for lower limb endovascular revascularization may be associated with increased risk of late mortality. OBJECTIVES: The purpose of this study was to determine whether PCDs are associated with all-cause mortality in a real-world setting. METHODS: DETECT is a nationwide, exhaustive retrospective cohort study using medico-administrative data from the French National Healthcare System representing >99% of the population. The main selection criterion was the first procedure of interest: endovascular revascularization for lower limb peripheral artery disease involving ≥1 balloon and/or stent performed between October 1, 2011, and December 31, 2019. Patients with or without PCDs were compared for all-cause mortality until December 31, 2021. RESULTS: A total of 259,137 patients were analyzed, with 20,083 (7.7%) treated with ≥1 PCD. After a median follow-up of 4.1 years (Q1-Q3: 2.3-6.4 years), a total of 5,385 deaths/73,923 person-years (PY) (7.3/100 PY) and 109,844 deaths/1,060,513 PY (10.4/100 PY) were observed in the PCD and control groups, respectively. After adjustment for confounding factors, PCD treatment was associated with a lower risk of mortality in multivariable Cox analyses (HR: 0.86; 95% CI: 0.84-0.89; P < 0.001). Similar results were observed using propensity score matching approaches based on either nearest-neighbor or exact matching. CONCLUSIONS: In a nationwide analysis based on large-scale real-world data, exposure to PCDs was not associated with a higher risk of mortality in patients undergoing endovascular revascularization for lower limb peripheral artery disease. (The DETECT Project; NCT05254106).

18th International HLA and Immunogenetics Workshop: Report on the SNP-HLA Reference Consortium (SHLARC) component.

The SNP-HLA Reference Consortium (SHLARC), a component of the 18th International HLA and Immunogenetics Workshop, is aimed at collecting diverse and extensive human leukocyte antigen (HLA) data to create custom reference panels and enhance HLA imputation techniques. Genome-wide association studies (GWAS) have significantly contributed to identifying genetic associations with various diseases. The HLA genomic region has emerged as the top locus in GWAS, particularly in immune-related disorders. However, the limited information provided by single nucleotide polymorphisms (SNPs), the hallmark of GWAS, poses challenges, especially in the HLA region, where strong linkage disequilibrium (LD) spans several megabases. HLA imputation techniques have been developed using statistical inference in response to these challenges. These techniques enable the prediction of HLA alleles from genotyped GWAS SNPs. Here we present the SHLARC activities, a collaborative effort to create extensive, and multi-ethnic reference panels to enhance HLA imputation accuracy.

Optimal population-specific HLA imputation with dimension reduction.

Human genomics has quickly evolved, powering genome-wide association studies (GWASs). SNP-based GWASs cannot capture the intense polymorphism of HLA genes, highly associated with disease susceptibility. There are methods to statistically impute HLA genotypes from SNP-genotypes data, but lack of diversity in reference panels hinders their performance. We evaluated the accuracy of the 1000 Genomes data as a reference panel for imputing HLA from admixed individuals of African and European ancestries, focusing on (a) the full dataset, (b) 10 replications from 6 populations, and (c) 19 conditions for the custom reference panels. The full dataset outperformed smaller models, with a good F1-score of 0.66 for HLA-B. However, custom models outperformed the multiethnic or population models of similar size (F1-scores up to 0.53, against up to 0.42). We demonstrated the importance of using genetically specific models for imputing populations, which are currently underrepresented in public datasets, opening the door to HLA imputation for every genetic population.

Applying Natural Language Processing to Textual Data From Clinical Data Warehouses: Systematic Review.

BACKGROUND: In recent years, health data collected during the clinical care process have been often repurposed for secondary use through clinical data warehouses (CDWs), which interconnect disparate data from different sources. A large amount of information of high clinical value is stored in unstructured text format. Natural language processing (NLP), which implements algorithms that can operate on massive unstructured textual data, has the potential to structure the data and make clinical information more accessible. OBJECTIVE: The aim of this review was to provide an overview of studies applying NLP to textual data from CDWs. It focuses on identifying the (1) NLP tasks applied to data from CDWs and (2) NLP methods used to tackle these tasks. METHODS: This review was performed according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. We searched for relevant articles in 3 bibliographic databases: PubMed, Google Scholar, and ACL Anthology. We reviewed the titles and abstracts and included articles according to the following inclusion criteria: (1) focus on NLP applied to textual data from CDWs, (2) articles published between 1995 and 2021, and (3) written in English. RESULTS: We identified 1353 articles, of which 194 (14.34%) met the inclusion criteria. Among all identified NLP tasks in the included papers, information extraction from clinical text (112/194, 57.7%) and the identification of patients (51/194, 26.3%) were the most frequent tasks. To address the various tasks, symbolic methods were the most common NLP methods (124/232, 53.4%), showing that some tasks can be partially achieved with classical NLP techniques, such as regular expressions or pattern matching that exploit specialized lexica, such as drug lists and terminologies. Machine learning (70/232, 30.2%) and deep learning (38/232, 16.4%) have been increasingly used in recent years, including the most recent approaches based on transformers. NLP methods were mostly applied to English language data (153/194, 78.9%). CONCLUSIONS: CDWs are central to the secondary use of clinical texts for research purposes. Although the use of NLP on data from CDWs is growing, there remain challenges in this field, especially with regard to languages other than English. Clinical NLP is an effective strategy for accessing, extracting, and transforming data from CDWs. Information retrieved with NLP can assist in clinical research and have an impact on clinical practice.

Strokecopilot: a literature-based clinical decision support system for acute ischemic stroke treatment.

BACKGROUND: Acute ischemic stroke (AIS) is an immediate emergency whose management is becoming more and more personalized while facing a limited number of neurologists with high expertise. Clinical decision support systems (CDSS) are digital tools leveraging information and artificial intelligence technologies. Here, we present the Strokecopilot project, a CDSS for the management of the acute phase of AIS. It has been designed to support the evidence-based medicine reasoning of neurologists regarding the indications of intravenous thrombolysis (IVT) and endovascular treatments (ET). METHODS: Reference populations were manually extracted from the field's main guidelines and randomized clinical trials (RCT). Their characteristics were harmonized in a computerized reference database. We developed a web application whose algorithm identifies the reference populations matching the patient's characteristics. It returns the latter's outcomes in a graphical user interface (GUI), whose design has been driven by real-world practices. RESULTS: Strokecopilot has been released at www.digitalneurology.net . The reference database includes 25 reference populations from 2 guidelines and 15 RCTs. After a request, the reference populations matching the patient characteristics are displayed with a summary and a meta-analysis of their results. The status regarding IVT and ET indications are presented as "in guidelines", "in literature", or "outside literature references". The GUI is updated to provide several levels of explanation. Strokecopilot may be updated as the literature evolves by loading a new version of the reference populations' database. CONCLUSION: Strokecopilot is a literature-based CDSS, developed to support neurologists in the management of the acute phase of AIS.

Multiple sclerosis: Exploring the limits and implications of genetic and environmental susceptibility.

OBJECTIVE: To explore and describe the basis and implications of genetic and environmental susceptibility to multiple sclerosis (MS) using the Canadian population-based data. BACKGROUND: Certain parameters of MS-epidemiology are directly observable (e.g., the recurrence-risk of MS in siblings and twins, the proportion of women among MS patients, the population-prevalence of MS, and the time-dependent changes in the sex-ratio). By contrast, other parameters can only be inferred from the observed parameters (e.g., the proportion of the population that is "genetically susceptible", the proportion of women among susceptible individuals, the probability that a susceptible individual will experience an environment "sufficient" to cause MS, and if they do, the probability that they will develop the disease). DESIGN/METHODS: The "genetically susceptible" subset (G) of the population (Z) is defined to include everyone with any non-zero life-time chance of developing MS under some environmental conditions. The value for each observed and non-observed epidemiological parameter is assigned a "plausible" range. Using both a Cross-sectional Model and a Longitudinal Model, together with established parameter relationships, we explore, iteratively, trillions of potential parameter combinations and determine those combinations (i.e., solutions) that fall within the acceptable range for both the observed and non-observed parameters. RESULTS: Both Models and all analyses intersect and converge to demonstrate that probability of genetic-susceptibitly, P(G), is limited to only a fraction of the population {i.e., P(G) ≤ 0.52)} and an even smaller fraction of women {i.e., P(G│F) < 0.32)}. Consequently, most individuals (particularly women) have no chance whatsoever of developing MS, regardless of their environmental exposure. However, for any susceptible individual to develop MS, requires that they also experience a "sufficient" environment. We use the Canadian data to derive, separately, the exponential response-curves for men and women that relate the increasing likelihood of developing MS to an increasing probability that a susceptible individual experiences an environment "sufficient" to cause MS. As the probability of a "sufficient" exposure increases, we define, separately, the limiting probability of developing MS in men (c) and women (d). These Canadian data strongly suggest that: (c < d ≤ 1). If so, this observation establishes both that there must be a "truly" random factor involved in MS pathogenesis and that it is this difference, rather than any difference in genetic or environmental factors, which primarily accounts for the penetrance difference between women and men. CONCLUSIONS: The development of MS (in an individual) requires both that they have an appropriate genotype (which is uncommon in the population) and that they have an environmental exposure "sufficient" to cause MS given their genotype. Nevertheless, the two principal findings of this study are that: P(G) ≤ 0.52)} and: (c < d ≤ 1). Threfore, even when the necessary genetic and environmental factors, "sufficient" for MS pathogenesis, co-occur for an individual, they still may or may not develop MS. Consequently, disease pathogenesis, even in this circumstance, seems to involve an important element of chance. Moreover, the conclusion that the macroscopic process of disease development for MS includes a "truly" random element, if replicated (either for MS or for other complex diseases), provides empiric evidence that our universe is non-deterministic.

Immunosuppressant drugs and quality-of-life outcomes in kidney transplant recipients: An international cohort study (EU-TRAIN).

Introduction: Patient-Reported Outcomes (PRO) integrate a wide range of holistic dimensions that arenot captured within clinical outcomes. Particularly, from induction treatment to maintenance therapy, patient quality-of-life (QoL) of kidney transplant recipients have been sparsely investigated in international settings. Methods: In a prospective, multi-centric cohort study, including nine transplant centers in four countries, we explored the QoL during the year following transplantation using validated elicitation instruments (EQ-5D-3L index with VAS) in a population of kidney transplant patients receiving immunosuppressive therapies. Calcineurin inhibitors (tacrolimus and ciclosporin), IMPD inhibitor (mycophenolate mofetil), and mTOR inhibitors (everolimus and sirolimus) were the standard-of-care (SOC) medications, together with tapering glucocorticoid therapy. We used EQ-5D and VAS data as QoL measures alongside descriptive statistics at inclusion, per country and hospital center. We computed the proportions of patients with different immunosuppressive therapy patterns, and using bivariate and multivariate analyses, assessed the variations of EQ-5D and VAS between baseline (i.e., inclusion Month 0) and follow up visits (Month 12). Results: Among 542 kidney transplant patients included and followed from November 2018 to June 2021, 491 filled at least one QoL questionnaire at least at baseline (Month 0). The majority of patients in all countries received tacrolimus and mycophenolate mofetil, ranging from 90.0% in Switzerland and Spain to 95.8% in Germany. At M12, a significant proportion of patients switched immunosuppressive drugs, with proportion varying from 20% in Germany to 40% in Spain and Switzerland. At visit M12, patients who kept SOC therapy had higher EQ-5D (by 8 percentage points, p < 0.05) and VAS (by 4 percentage points, p < 0.1) scores than switchers. VAS scores were generally lower than EQ-5D (mean 0.68 [0.5-0.8] vs. 0.85 [0.8-1]). Discussion: Although overall a positive trend in QoL was observed, the formal analyses did not show any significant improvements in EQ-5D scores or VAS. Only when the effect of a therapy use was separated from the effect of switching, the VAS score was significantly worse for switchers during the follow up period, irrespective of the therapy type. If adjusted for patient characteristics and medical history (e.g., gender, BMI, eGRF, history of diabetes), VAS and EQ-5D delivered sound PRO measures for QoL assessments during the year following renal transplantation.

Patient-centric synthetic data generation, no reason to risk re-identification in biomedical data analysis.

While nearly all computational methods operate on pseudonymized personal data, re-identification remains a risk. With personal health data, this re-identification risk may be considered a double-crossing of patients' trust. Herein, we present a new method to generate synthetic data of individual granularity while holding on to patients' privacy. Developed for sensitive biomedical data, the method is patient-centric as it uses a local model to generate random new synthetic data, called an "avatar data", for each initial sensitive individual. This method, compared with 2 other synthetic data generation techniques (Synthpop, CT-GAN), is applied to real health data with a clinical trial and a cancer observational study to evaluate the protection it provides while retaining the original statistical information. Compared to Synthpop and CT-GAN, the Avatar method shows a similar level of signal maintenance while allowing to compute additional privacy metrics. In the light of distance-based privacy metrics, each individual produces an avatar simulation that is on average indistinguishable from 12 other generated avatar simulations for the clinical trial and 24 for the observational study. Data transformation using the Avatar method both preserves, the evaluation of the treatment's effectiveness with similar hazard ratios for the clinical trial (original HR = 0.49 [95% CI, 0.39-0.63] vs. avatar HR = 0.40 [95% CI, 0.31-0.52]) and the classification properties for the observational study (original AUC = 99.46 (s.e. 0.25) vs. avatar AUC = 99.84 (s.e. 0.12)). Once validated by privacy metrics, anonymous synthetic data enable the creation of value from sensitive pseudonymized data analyses by tackling the risk of a privacy breach.

KiT-GENIE, the French genetic biobank of kidney transplantation.

KiT-GENIE is a monocentric DNA biobank set up to consolidate the very rich and homogeneous DIVAT French cohort of kidney donors and recipients (D/R) in order to explore the molecular factors involved in kidney transplantation outcomes. We collected DNA samples for kidney transplantations performed in Nantes, and we leveraged GWAS genotyping data for securing high-quality genetic data with deep SNP and HLA annotations through imputations and for inferring D/R genetic ancestry. Overall, the biobank included 4217 individuals (n = 1945 D + 2,272 R, including 1969 D/R pairs), 7.4 M SNPs and over 200 clinical variables. KiT-GENIE represents an accurate snapshot of kidney transplantation clinical practice in Nantes between 2002 and 2018, with an enrichment in living kidney donors (17%) and recipients with focal segmental glomerulosclerosis (4%). Recipients were predominantly male (63%), of European ancestry (93%), with a mean age of 51yo and 86% experienced their first graft over the study period. D/R pairs were 93% from European ancestry, and 95% pairs exhibited at least one HLA allelic mismatch. The mean follow-up time was 6.7 years with a hindsight up to 25 years. Recipients experienced biopsy-proven rejection and graft loss for 16.6% and 21.3%, respectively. KiT-GENIE constitutes one of the largest kidney transplantation genetic cohorts worldwide to date. It includes homogeneous high-quality clinical and genetic data for donors and recipients, hence offering a unique opportunity to investigate immunogenetic and genetic factors, as well as donor-recipient interactions and mismatches involved in rejection, graft survival, primary disease recurrence and other comorbidities.

Immune Profiling Reveals the T-Cell Effect of Ocrelizumab in Early Relapsing-Remitting Multiple Sclerosis.

BACKGROUND AND OBJECTIVES: Ocrelizumab (OCR), a humanized anti-CD20 monoclonal antibody, is highly efficient in patients with relapsing-remitting multiple sclerosis (RR-MS). We assessed early cellular immune profiles and their association with disease activity at treatment start and under therapy, which may provide new clues on the mechanisms of action of OCR and on the disease pathophysiology. METHODS: A first group of 42 patients with an early RR-MS, never exposed to disease-modifying therapy, was included in 11 centers participating to an ancillary study of the ENSEMBLE trial (NCT03085810) to evaluate the effectiveness and safety of OCR. The phenotypic immune profile was comprehensively assessed by multiparametric spectral flow cytometry at baseline and after 24 and 48 weeks of OCR treatment on cryopreserved peripheral blood mononuclear cells and analyzed in relation to disease clinical activity. A second group of 13 untreated patients with RR-MS was included for comparative analysis of peripheral blood and CSF. The transcriptomic profile was assessed by single-cell qPCRs of 96 genes of immunologic interest. RESULTS: Using an unbiased analysis, we found that OCR as an effect on 4 clusters of CD4+ T cells: one corresponding to naive CD4+ T cells was increased, the other clusters corresponded to effector memory (EM) CD4+CCR6- T cells expressing homing and migration markers, 2 of them also expressing CCR5 and were decreased by the treatment. Of interest, one CD8+ T-cell cluster was decreased by OCR corresponding to EM CCR5-expressing T cells with high expression of the brain homing markers CD49d and CD11a and correlated with the time elapsed since the last relapse. These EM CD8+CCR5+ T cells were enriched in the CSF of patients with RR-MS and corresponded to activated and cytotoxic cells. DISCUSSION: Our study provides novel insights into the mode of action of anti-CD20, pointing toward the role of EM T cells, particularly a subset of CD8 T cells expressing CCR5.

Identification of Novel Genetic Risk Factors for Focal Segmental Glomerulosclerosis in Children: Results From the Chronic Kidney Disease in Children (CKiD) Cohort.

RATIONALE & OBJECTIVE: Focal segmental glomerulosclerosis (FSGS) is a major cause of pediatric nephrotic syndrome, and African Americans exhibit an increased risk for developing FSGS compared with other populations. Predisposing genetic factors have previously been described in adults. Here we performed genomic screening of primary FSGS in a pediatric African American population. STUDY DESIGN: Prospective cohort with case-control genetic association study design. SETTING & PARTICIPANTS: 140 African American children with chronic kidney disease from the Chronic Kidney Disease in Children (CKiD) cohort, including 32 cases with FSGS. PREDICTORS: Over 680,000 common single-nucleotide polymorphisms (SNPs) were tested for association. We also ran a pathway enrichment analysis and a human leucocyte antigen (HLA)-focused association study. OUTCOME: Primary biopsy-proven pediatric FSGS. ANALYTICAL APPROACH: Multivariate logistic regression models. RESULTS: The genome-wide association study revealed 169 SNPs from 14 independent loci significantly associated with FSGS (false discovery rate [FDR]<5%). We observed notable signals for genetic variants within the APOL1 (P=8.6×10-7; OR, 25.8 [95% CI, 7.1-94.0]), ALMS1 (P=1.3×10-7; 13.0% in FSGS cases vs 0% in controls), and FGFR4 (P=4.3×10-6; OR, 24.8 [95% CI, 6.3-97.7]) genes, all of which had previously been associated with adult FSGS, kidney function, or chronic kidney disease. We also highlighted novel, functionally relevant genes, including GRB2 (which encodes a slit diaphragm protein promoting podocyte structure through actin polymerization) and ITGB1 (which is linked to renal injuries). Our results suggest a major role for immune responses and antigen presentation in pediatric FSGS through (1) associations with SNPs in PTPRJ (or CD148, P=3.5×10-7), which plays a role in T-cell receptor signaling, (2) HLA-DRB1∗11:01 association (P=6.1×10-3; OR, 4.5 [95% CI, 1.5-13.0]), and (3) signaling pathway enrichment (P=1.3×10-6). LIMITATIONS: Sample size and no independent replication cohort with genomic data readily available. CONCLUSIONS: Our genetic study has identified functionally relevant risk factors and the importance of immune regulation for pediatric primary FSGS, which contributes to a better description of its molecular pathophysiological mechanisms. PLAIN-LANGUAGE SUMMARY: We assessed the genetic risk factors for primary focal segmental glomerulosclerosis (FSGS) by simultaneously testing over 680,000 genetic markers spread across the genome in 140 children, including 32 with FSGS lesions. Fourteen independent genetic regions were significantly associated with pediatric FSGS, including APOL1 and ALMS1-NAT8, which were previously found to be associated with FSGS and chronic kidney diseases in adults. Novel genes with relevant biological functions were also highlighted, such as GRB2 and FGFR4, which play a role in the kidney filtration barrier and in kidney cell differentiation, respectively. Finally, we revealed the importance of immune regulation in pediatric FSGS through associations involving cell surface proteins presenting antigens to the immune system and interacting with T-cell receptors.

Impact of the duration of oestradiol treatment on live birth rate in Hormonal Replacement Therapy cycle before frozen blastocyst transfer.

Although the duration of progesterone administration in Hormonal Replacement Therapy (HRT) cycles before frozen embryo transfer is standardized, the optimal duration of oestrogen treatment remains controversial. In this monocentric retrospective study conducted in all single frozen blastocyst transfer (FBT) performed with HRT between January 2016 and July 2019, we evaluated the association between the duration of oestradiol treatment before FBT and live birth rate (LBR) in HRT cycles. Cycles were gathered in 3 groups according to quartiles of duration of oestrogen treatment. LBR was compared across the 3 groups and multivariate analysis was performed. We included 2235 single FBT cycles; 507, 1257 and 471 with E2 treatment below 23 days, 23-30 days (reference) and more than 30 days respectively. After multivariate analysis and adjustment, no significant difference in LBR was found between below 23 or more than 30 days and reference groups (OR = 0.93 [0.68-1.27] and OR = 1.29 [0.88-1.89] respectively). Complementary sensitivity analysis led to a non-significant adjusted OR = 1.66 [IC 0.9-3.1]. In conclusion, our study showed that the duration of E2 treatment in HRT cycles before FBT is not associated with LBR.

Semi-supervised clustering of quaternion time series: Application to gait analysis in multiple sclerosis using motion sensor data.

Recent approaches in gait analysis involve the use of wearable motion sensors to extract spatio-temporal parameters that characterize multiple aspects of an individual's gait. In particular, the medical community could largely benefit from this type of devices as they could provide the clinicians with a valuable tool for assessing gait impairment. Motion sensor data are however complex and there is an urgent unmet need to develop sound statistical methods for analyzing such data and extracting clinically relevant information. In this article, we measure gait by following the hip rotation over time and the resulting statistical unit is a time series of unit quaternions. We explore the possibility to form groups of patients with similar walking impairment by taking into account their walking data and their global decease severity with semi-supervised clustering. We generalize a compromise-based method named hclustcompro to unit quaternion time series by combining it with the proper dissimilarity quaternion dynamic time warping. We apply this method on patients diagnosed with multiple sclerosis to form groups of patients with similar walking deficiencies while accounting for the clinical assessment of their overall disability. We also compare the compromise-based clustering approach with the method mergeTrees that falls into a sub-class of ensemble clustering named collaborative clustering. The results provide a first proof of both the interest of using wearable motion sensors for assessing gait impairment and the use of prior knowledge to guide the clustering process. It also demonstrates that compromise-based clustering is a more appropriate approach in this context.

Multiple sclerosis clinical decision support system based on projection to reference datasets.

OBJECTIVE: Multiple sclerosis (MS) is a multifactorial disease with increasingly complicated management. Our objective is to use on-demand computational power to address the challenges of dynamically managing MS. METHODS: A phase 3 clinical trial data (NCT00906399) were used to contextualize the medication efficacy of peg-interferon beta-1a vs placebo on patients with relapsing-remitting MS (RRMS). Using a set of reference patients (PORs), selected based on adequate features similar to those of an individual patient, we visualize disease activity by measuring the percentage of relapses, accumulation of new T2 lesions on MRI, and worsening EDSS during the clinical trial. RESULTS: We developed MS Vista, a functional prototype of clinical decision support system (CDSS), with a user-centered design and distributed infrastructure. MS Vista shows the medication efficacy of peginterferon beta-1a versus placebo for each individual patient with RRMS. In addition, MS Vista initiated the integration of a longitudinal magnetic resonance imaging (MRI) viewer and interactive dual physician-patient data display to facilitate communication. INTERPRETATION: The pioneer use of PORs for each individual patient enables personalized analytics sustaining the dialog between neurologists, patients and caregivers with quantified evidence.

Addressing the Challenges and Barriers to the Integration of Machine Learning into Clinical Practice: An Innovative Method to Hybrid Human-Machine Intelligence.

Machine learning (ML) models have proven their potential in acquiring and analyzing large amounts of data to help solve real-world, complex problems. Their use in healthcare is expected to help physicians make diagnoses, prognoses, treatment decisions, and disease outcome predictions. However, ML solutions are not currently deployed in most healthcare systems. One of the main reasons for this is the provenance, transparency, and clinical utility of the training data. Physicians reject ML solutions if they are not at least based on accurate data and do not clearly include the decision-making process used in clinical practice. In this paper, we present a hybrid human-machine intelligence method to create predictive models driven by clinical practice. We promote the use of quality-approved data and the inclusion of physician reasoning in the ML process. Instead of training the ML algorithms on the given data to create predictive models (conventional method), we propose to pre-categorize the data according to the expert physicians' knowledge and experience. Comparing the results of the conventional method of ML learning versus the hybrid physician-algorithm method showed that the models based on the latter can perform better. Physicians' engagement is the most promising condition for the safe and innovative use of ML in healthcare.

Linking Biomedical Data Warehouse Records With the National Mortality Database in France: Large-scale Matching Algorithm.

BACKGROUND: Often missing from or uncertain in a biomedical data warehouse (BDW), vital status after discharge is central to the value of a BDW in medical research. The French National Mortality Database (FNMD) offers open-source nominative records of every death. Matching large-scale BDWs records with the FNMD combines multiple challenges: absence of unique common identifiers between the 2 databases, names changing over life, clerical errors, and the exponential growth of the number of comparisons to compute. OBJECTIVE: We aimed to develop a new algorithm for matching BDW records to the FNMD and evaluated its performance. METHODS: We developed a deterministic algorithm based on advanced data cleaning and knowledge of the naming system and the Damerau-Levenshtein distance (DLD). The algorithm's performance was independently assessed using BDW data of 3 university hospitals: Lille, Nantes, and Rennes. Specificity was evaluated with living patients on January 1, 2016 (ie, patients with at least 1 hospital encounter before and after this date). Sensitivity was evaluated with patients recorded as deceased between January 1, 2001, and December 31, 2020. The DLD-based algorithm was compared to a direct matching algorithm with minimal data cleaning as a reference. RESULTS: All centers combined, sensitivity was 11% higher for the DLD-based algorithm (93.3%, 95% CI 92.8-93.9) than for the direct algorithm (82.7%, 95% CI 81.8-83.6; P<.001). Sensitivity was superior for men at 2 centers (Nantes: 87%, 95% CI 85.1-89 vs 83.6%, 95% CI 81.4-85.8; P=.006; Rennes: 98.6%, 95% CI 98.1-99.2 vs 96%, 95% CI 94.9-97.1; P<.001) and for patients born in France at all centers (Nantes: 85.8%, 95% CI 84.3-87.3 vs 74.9%, 95% CI 72.8-77.0; P<.001). The DLD-based algorithm revealed significant differences in sensitivity among centers (Nantes, 85.3% vs Lille and Rennes, 97.3%, P<.001). Specificity was >98% in all subgroups. Our algorithm matched tens of millions of death records from BDWs, with parallel computing capabilities and low RAM requirements. We used the Inseehop open-source R script for this measurement. CONCLUSIONS: Overall, sensitivity/recall was 11% higher using the DLD-based algorithm than that using the direct algorithm. This shows the importance of advanced data cleaning and knowledge of a naming system through DLD use. Statistically significant differences in sensitivity between groups could be found and must be considered when performing an analysis to avoid differential biases. Our algorithm, originally conceived for linking a BDW with the FNMD, can be used to match any large-scale databases. While matching operations using names are considered sensitive computational operations, the Inseehop package released here is easy to run on premises, thereby facilitating compliance with cybersecurity local framework. The use of an advanced deterministic matching algorithm such as the DLD-based algorithm is an insightful example of combining open-source external data to improve the usage value of BDWs.