Enhanced quantitation of pathological α-synuclein in patient biospecimens by RT-QuIC seed amplification assays.

Disease associated pathological aggregates of alpha-synuclein (αSynD) exhibit prion-like spreading in synucleinopathies such as Parkinson's disease (PD) and dementia with Lewy bodies (DLB). Seed amplification assays (SAAs) such as real-time quaking-induced conversion (RT-QuIC) have shown high diagnostic sensitivity and specificity for detecting proteopathic αSynD seeds in a variety of biospecimens from PD and DLB patients. However, the extent to which relative proteopathic seed concentrations are useful as indices of a patient's disease stage or prognosis remains unresolved. One feature of current SAAs that complicates attempts to correlate SAA results with patients' clinical and other laboratory findings is their quantitative imprecision, which has typically been limited to discriminating large differences (e.g. 5-10 fold) in seed concentration. We used end-point dilution (ED) RT-QuIC assays to determine αSynD seed concentrations in patient biospecimens and tested the influence of various assay variables such as serial dilution factor, replicate number and data processing methods. The use of 2-fold versus 10-fold dilution factors and 12 versus 4 replicate reactions per dilution reduced ED-RT-QuIC assay error by as much as 70%. This enhanced assay format discriminated as little as 2-fold differences in αSynD seed concentration besides detecting ~2-16-fold seed reductions caused by inactivation treatments. In some scenarios, analysis of the data using Poisson and midSIN algorithms provided more consistent and statistically significant discrimination of different seed concentrations. We applied our improved assay strategies to multiple diagnostically relevant PD and DLB antemortem patient biospecimens, including cerebrospinal fluid, skin, and brushings of the olfactory mucosa. Using ED αSyn RT-QuIC as a model SAA, we show how to markedly improve the inter-assay reproducibility and quantitative accuracy. Enhanced quantitative SAA accuracy should facilitate assessments of pathological seeding activities as biomarkers in proteinopathy diagnostics and prognostics, as well as in patient cohort selection and assessments of pharmacodynamics and target engagement in drug trials.

Exploring associations between viral titer measurements and disease outcomes in ferrets inoculated with 125 contemporary influenza A viruses.

As use of the ferret model to study influenza A virus (IAV) pathogenicity increases, periodic assessment of data generated in this model is warranted, to identify features associated with virus replication throughout the respiratory tract and to refine future analyses. However, protocol-specific differences present between independent laboratories limit easy aggregation of virological data. We compiled viral titer and clinical data from >1,000 ferrets inoculated with 125 contemporary IAV under a consistent experimental protocol (including high- and low-pathogenicity avian, swine-origin, and human viruses, spanning H1, H2, H3, H5, H7, and H9 subtypes) and examined which meaningful and statistically supported associations were present among numerous quantitative measurements. Viral titers correlated positively between ferret nasal turbinate tissue, lung tissue, and nasal wash specimens, though the strength of the associations varied, notably regarding the particular nasal wash summary measure employed and properties of the virus itself. Use of correlation coefficients and mediation analyses further supported the interconnectedness of viral titer measurements taken at different sites throughout the respiratory tract. IAV possessing mammalian host adaptation markers in the HA and PB2 exhibited more rapid growth in the ferret upper respiratory tract early after infection, supported by quantities derived from infectious titer data to capture infection progression, compared with viruses bearing hallmarks of avian IAV. Collectively, this work identifies summary metrics most closely linked with virological and phenotypic outcomes in ferrets, supporting continued refinement of data analyzed from in vivo experimentation, notably from studies conducted to evaluate the public health risk posed by novel and emerging IAV.IMPORTANCEFerrets are frequently employed to study the pandemic potential of novel and emerging influenza A viruses. However, systematic retrospective analyses of data generated from these experiments are rarely performed, limiting our ability to identify trends in this data and explore how analyses can be refined. Using logarithmic viral titer and clinical data aggregated from one research group over 20 years, we assessed which meaningful and statistically supported associations were present among numerous quantitative measurements obtained from influenza A virus (IAV)-infected ferrets, including those capturing viral titers, infection progression, and disease severity. We identified numerous linear correlations between parameters assessing virus replication at discrete sites in vivo, including parameters capturing infection progression not frequently employed in the field, and sought to investigate the interconnected nature of these associations. This work supports continued refinement of data analyzed from in vivo experimentation, notably from studies which evaluate the public health risk posed by IAV.

Apolipoprotein A1 and high-density lipoprotein limit low-density lipoprotein transcytosis by binding SR-B1.

Atherosclerosis results from the deposition and oxidation of LDL and immune cell infiltration in the sub-arterial space leading to arterial occlusion. Studies have shown that transcytosis transports circulating LDL across endothelial cells lining blood vessels. LDL transcytosis is initiated by binding to either scavenger receptor B1 (SR-B1) or activin A receptor-like kinase 1 on the apical side of endothelial cells leading to its transit and release on the basolateral side. HDL is thought to partly protect individuals from atherosclerosis due to its ability to remove excess cholesterol and act as an antioxidant. Apolipoprotein A1 (APOA1), an HDL constituent, can bind to SR-B1, raising the possibility that APOA1/HDL can compete with LDL for SR-B1 binding, thereby limiting LDL deposition in the sub-arterial space. To examine this possibility, we used in vitro approaches to quantify the internalization and transcytosis of fluorescent LDL in coronary endothelial cells. Using microscale thermophoresis and affinity capture, we find that SR-B1 and APOA1 interact and that binding is enhanced when using the cardioprotective variant of APOA1 termed Milano (APOA1-Milano). In male mice, transiently increasing the levels of HDL reduced the acute deposition of fluorescently labeled LDL in the atheroprone inner curvature of the aorta. Reduced LDL deposition was also observed when increasing circulating wild-type APOA1 or the APOA1-Milano variant, with a more robust inhibition from the APOA1-Milano. The results suggest that HDL may limit SR-B1-mediated LDL transcytosis and deposition, adding to the mechanisms by which it can act as an atheroprotective particle.

Utility of Human In Vitro Data in Risk Assessments of Influenza A Virus Using the Ferret Model.

As influenza A viruses (IAV) continue to cross species barriers and cause human infection, the establishment of risk assessment rubrics has improved pandemic preparedness efforts. In vivo pathogenicity and transmissibility evaluations in the ferret model represent a critical component of this work. As the relative contribution of in vitro experimentation to these rubrics has not been closely examined, we sought to evaluate to what extent viral titer measurements over the course of in vitro infections are predictive or correlates of nasal wash and tissue measurements for IAV infections in vivo. We compiled data from ferrets inoculated with an extensive panel of over 50 human and zoonotic IAV (inclusive of swine-origin and high- and low-pathogenicity avian influenza viruses associated with human infection) under a consistent protocol, with all viruses concurrently tested in a human bronchial epithelial cell line (Calu-3). Viral titers in ferret nasal wash specimens and nasal turbinate tissue correlated positively with peak titer in Calu-3 cells, whereas additional phenotypic and molecular determinants of influenza virus virulence and transmissibility in ferrets varied in their association with in vitro viral titer measurements. Mathematical modeling was used to estimate more generalizable key replication kinetic parameters from raw in vitro viral titers, revealing commonalities between viral infection progression in vivo and in vitro. Meta-analyses inclusive of IAV that display a diverse range of phenotypes in ferrets, interpreted with mathematical modeling of viral kinetic parameters, can provide critical information supporting a more rigorous and appropriate contextualization of in vitro experiments toward pandemic preparedness. IMPORTANCE Both in vitro and in vivo models are employed for assessing the pandemic potential of novel and emerging influenza A viruses in laboratory settings, but systematic examinations of how well viral titer measurements obtained in vitro align with results from in vivo experimentation are not frequently performed. We show that certain viral titer measurements following infection of a human bronchial epithelial cell line are positively correlated with viral titers in specimens collected from virus-inoculated ferrets and employ mathematical modeling to identify commonalities between viral infection progression between both models. These analyses provide a necessary first step in enhanced interpretation and incorporation of in vitro-derived data in risk assessment activities and highlight the utility of employing mathematical modeling approaches to more closely examine features of virus replication not identifiable by experimental studies alone.

Correction: Time to revisit the endpoint dilution assay and to replace the TCID50 as a measure of a virus sample's infection concentration.

[This corrects the article DOI: 10.1371/journal.pcbi.1009480.].

Evaluating the economic burden of acute myeloid leukemia in Canada.

Background Acute myeloid leukemia (AML) represents a significant burden for patients and their families, and to the healthcare system. This study estimated the total cost of illness associated with newly diagnosed AML patients in Canada. Methods The economic burden of AML was estimated using an incidence-based model, analyzing different types of AML cases in Canada. Direct and indirect costs were calculated using scientific literature and Canadian clinical experts' inputs. Patients were categorized depending on their eligibility for intensive chemotherapy (fit and unfit patients) as well as according to age and cytogenetic markers. Results The total average cost of AML per patient is estimated to be $178,073 with a cost of $210,983 and $145,163 for fit and unfit patients, respectively. The costs related to treatment represent half of the total average cost (52%), followed by hematopoietic stem cell transplant (23%), best supportive care (16%), productivity loss (6%) and wastage (4%) Conclusions For patients with AML, the costs associated with fit patients are higher than unfit patients. Hospitalization and best supportive care costs are key cost drivers for the total costs of fit and unfit patients, respectively. This study highlights that AML is associated with a significant economic burden in Canada.

Correction: A mathematical model describing the localization and spread of influenza A virus infection within the human respiratory tract.

[This corrects the article DOI: 10.1371/journal.pcbi.1007705.].

Time to revisit the endpoint dilution assay and to replace the TCID50 as a measure of a virus sample's infection concentration.

The endpoint dilution assay's output, the 50% infectious dose (ID50), is calculated using the Reed-Muench or Spearman-Kärber mathematical approximations, which are biased and often miscalculated. We introduce a replacement for the ID50 that we call Specific INfection (SIN) along with a free and open-source web-application, midSIN (https://midsin.physics.ryerson.ca) to calculate it. midSIN computes a virus sample's SIN concentration using Bayesian inference based on the results of a standard endpoint dilution assay, and requires no changes to current experimental protocols. We analyzed influenza and respiratory syncytial virus samples using midSIN and demonstrated that the SIN/mL reliably corresponds to the number of infections a sample will cause per mL. It can therefore be used directly to achieve a desired multiplicity of infection, similarly to how plaque or focus forming units (PFU, FFU) are used. midSIN's estimates are shown to be more accurate and robust than the Reed-Muench and Spearman-Kärber approximations. The impact of endpoint dilution plate design choices (dilution factor, replicates per dilution) on measurement accuracy is also explored. The simplicity of SIN as a measure and the greater accuracy provided by midSIN make them an easy and superior replacement for the TCID50 and other in vitro culture ID50 measures. We hope to see their universal adoption to measure the infectivity of virus samples.

A mathematical model describing the localization and spread of influenza A virus infection within the human respiratory tract.

Within the human respiratory tract (HRT), virus diffuses through the periciliary fluid (PCF) bathing the epithelium. But virus also undergoes advection: as the mucus layer sitting atop the PCF is pushed along by the ciliated cell's beating cilia, the PCF and its virus content are also pushed along, upwards towards the nose and mouth. While many mathematical models (MMs) have described the course of influenza A virus (IAV) infections in vivo, none have considered the impact of both diffusion and advection on the kinetics and localization of the infection. The MM herein represents the HRT as a one-dimensional track extending from the nose down towards the lower HRT, wherein stationary cells interact with IAV which moves within (diffusion) and along with (advection) the PCF. Diffusion was found to be negligible in the presence of advection which effectively sweeps away IAV, preventing infection from disseminating below the depth at which virus first deposits. Higher virus production rates (10-fold) are required at higher advection speeds (40 µm/s) to maintain equivalent infection severity and timing. Because virus is entrained upwards, upper parts of the HRT see more virus than lower parts. As such, infection peaks and resolves faster in the upper than in the lower HRT, making it appear as though infection progresses from the upper towards the lower HRT, as reported in mice. When the spatial MM is expanded to include cellular regeneration and an immune response, it reproduces tissue damage levels reported in patients. It also captures the kinetics of seasonal and avian IAV infections, via parameter changes consistent with reported differences between these strains, enabling comparison of their treatment with antivirals. This new MM offers a convenient and unique platform from which to study the localization and spread of respiratory viral infections within the HRT.

Correction: A mathematical model describing the localization and spread of influenza A virus infection within the human respiratory tract.

[This corrects the article DOI: 10.1371/journal.pcbi.1007705.].

Quantification of Ebola virus replication kinetics in vitro.

Mathematical modelling has successfully been used to provide quantitative descriptions of many viral infections, but for the Ebola virus, which requires biosafety level 4 facilities for experimentation, modelling can play a crucial role. Ebola virus modelling efforts have primarily focused on in vivo virus kinetics, e.g., in animal models, to aid the development of antivirals and vaccines. But, thus far, these studies have not yielded a detailed specification of the infection cycle, which could provide a foundational description of the virus kinetics and thus a deeper understanding of their clinical manifestation. Here, we obtain a diverse experimental data set of the Ebola virus infection in vitro, and then make use of Bayesian inference methods to fully identify parameters in a mathematical model of the infection. Our results provide insights into the distribution of time an infected cell spends in the eclipse phase (the period between infection and the start of virus production), as well as the rate at which infectious virions lose infectivity. We suggest how these results can be used in future models to describe co-infection with defective interfering particles, which are an emerging alternative therapeutic.

Canadian economic impact of improved workplace productivity in patients with major depressive disorder treated with vortioxetine.

OBJECTIVE: The AtWoRC study is an interventional, open-label Canadian study that demonstrated significant improvements in cognitive function and workplace productivity in patients with major depressive disorder (MDD) treated with vortioxetine for a current major depressive episode. The objective of the present analysis was to assess the Canadian economic impact of improved workplace productivity based on the AtWoRC study results. METHODS: The economic impact of improved productivity in patients with MDD treated with vortioxetine was assessed over a 52-week period considering productivity loss due to absenteeism and presenteeism using the standard human capital approach and an employer's perspective. Absenteeism was measured with the Work Productivity and Activity Impairment questionnaire; and presenteeism with the Work Limitation Questionnaire. Productivity gains following treatment initiation with vortioxetine were estimated using the difference from baseline. RESULTS: In the AtWoRC study, patients at baseline reportedly missed, in the past 7 days, an average of 8.1 h due to absenteeism and 3.0 h due to presenteeism. Following 52 weeks of treatment with vortioxetine, patients reportedly missed an average of 4.9 h due to absenteeism and 2.0 h due to presenteeism. This improved workplace productivity translated into savings of C$110.64 for 1 week of work following 52 weeks of treatment. The cumulative 52-week economic impact showed potential savings of C$4,550 when factoring in the cost of therapy. CONCLUSION: This study suggested that workplace productivity gain due to an improvement in symptoms of MDD following treatment with vortioxetine will lead to substantial cost savings for the Canadian economy.

Real-World Experience With Apremilast in the Treatment of Adults With Moderate to Severe Plaque Psoriasis in Québec: A Claims-Based Analysis of Drug Utilization and Healthcare Resource Utilization.

BACKGROUND: In Québec, targeted biologic therapies for moderate to severe plaque psoriasis are restricted to patients who have not responded to phototherapy or conventional systemic treatment, primarily due to high drug costs. Apremilast, an oral treatment for plaque psoriasis, was added to the Québec provincial health insurance plan (Régie de l'assurance maladie du Québec; RAMQ) formulary in 2015, making this the only province in Canada with public drug plan reimbursement for apremilast. OBJECTIVES: The aim of this study is to describe patients' characteristics, treatment patterns, healthcare resource utilization (HCRU), and associated costs and to measure real-world budget impact of using apremilast before biologics in plaque psoriasis. METHODS: This study was performed using RAMQ drug claims and medical services data. Patients diagnosed with psoriasis between January 2015 and December 2017 were identified. Medical services and prescription claims were categorized as all-cause and psoriasis-related. Using RAMQ database estimates, a 3-year budget impact analysis was developed comparing treatment cost with and without the addition of apremilast to the formulary. RESULTS: In all, 540 patients were identified (apremilast: n = 92; biologics: n = 448). Comorbidity burden and treatment persistence and adherence were comparable between apremilast and biologic users. The year following the index date, all-cause HCRU was lower for apremilast versus biologic users (CAN$19 763 vs CAN$28 025; P < .01), mainly driven by drug cost. Using apremilast before biologics resulted in an estimated RAMQ net savings of CAN$49 290 (2015), CAN$746 856 (2016), and CAN$1 216 512 (2017), and a total savings of CAN$2 012 658 since apremilast's addition to the formulary. CONCLUSION: Adding apremilast to the drug formulary of other Canadian provinces could result in significant healthcare savings.

Cost-Utility Analysis of Imaging for Surveillance and Diagnosis of Hepatocellular Carcinoma.

OBJECTIVE. The purpose of this study is to compare imaging-based surveillance and diagnostic strategies in patients at risk for hepatocellular carcinoma (HCC) while taking into account technically inadequate examinations and patient compliance. MATERIALS AND METHODS. A Markov model simulated seven strategies for HCC surveillance and diagnosis in patients with cirrhosis: strategy A, ultrasound (US) for surveillance and CT for diagnosis; strategy B, US for surveillance and complete MRI for diagnosis; strategy C, US for surveillance and CT for inadequate or positive surveillance; strategy D, US for surveillance and complete MRI for inadequate or positive surveillance; strategy E, surveillance and diagnosis with CT followed by complete MRI for inadequate surveillance; strategy F, surveillance and diagnosis with complete MRI followed by CT for inadequate surveillance; and strategy G, surveillance with abbreviated MRI followed by CT for inadequate surveillance or complete MRI for positive surveillance. Two compliance scenarios were evaluated: optimal and conservative. For each scenario, the most cost-effective strategy was based on a willingness-to-pay threshold of $50,000 (Canadian) per quality-adjusted life year (QALY). Sensitivity analyses were performed. RESULTS. Base-case analysis revealed that strategy E was the most cost-effective when compliance was optimal ($13,631/QALY), and strategy G was the most cost-effective when compliance was conservative ($39,681/QALY). Sensitivity analyses supported the base-case analysis in the optimal compliance scenario, but several parameters altered the most cost-effective strategy in the conservative compliance scenario. CONCLUSION. In an optimal compliance scenario, CT for HCC surveillance and diagnosis and complete MRI for inadequate CT was most cost-effective. In a conservative compliance scenario, abbreviated MRI may be an alternative to US-based surveillance.

Modelling endurance and resumption times for repetitive one-hand pushing.

This study's objective was to develop models of endurance time (ET), as a function of load level (LL), and of resumption time (RT) after loading as a function of both LL and loading time (LT) for repeated loadings. Ten male participants with experience in construction work each performed 15 different one-handed repetaed pushing tasks at shoulder height with varied exerted force and duration. These data were used to create regression models predicting ET and RT. It is concluded that power law relationships are most appropriate to use when modelling ET and RT. While the data the equations are based on are limited regarding number of participants, gender, postures, magnitude and type of exerted force, the paper suggests how this kind of modelling can be used in job design and in further research. Practitioner Summary: Adequate muscular recovery during work-shifts is important to create sustainable jobs. This paper describes mathematical modelling and presents models for endurance times and resumption times (an aspect of recovery need), based on data from an empirical study. The models can be used to help manage fatigue levels in job design.

A drug-disease model describing the effect of oseltamivir neuraminidase inhibition on influenza virus progression.

A population drug-disease model was developed to describe the time course of influenza virus with and without oseltamivir treatment and to investigate opportunities for antiviral combination therapy. Data included viral titers from 208 subjects, across 4 studies, receiving placebo and oseltamivir at 20 to 200 mg twice daily for 5 days. A 3-compartment mathematical model, comprising target cells infected at rate ß, free virus produced at rate p and cleared at rate c, and infected cells cleared at rate δ, was implemented in NONMEM with an inhibitory Hill function on virus production (p), accounting for the oseltamivir effect. In congruence with clinical data, the model predicts that the standard 75-mg regimen initiated 2 days after infection decreased viral shedding duration by 1.5 days versus placebo; the 150-mg regimen decreased shedding by an additional average 0.25 day. The model also predicts that initiation of oseltamivir sooner postinfection, specifically at day 0.5 or 1, results in proportionally greater decreases in viral shedding duration of 5 and 3.5 days, respectively. Furthermore, the model suggests that combining oseltamivir (acting to subdue virus production rate) with an antiviral whose activity decreases viral infectivity (ß) results in a moderate additive effect dependent on therapy initiation time. In contrast, the combination of oseltamivir with an antiviral whose activity increases viral clearance (c) shows significant additive effects independent of therapy initiation time. The utility of the model for investigating the pharmacodynamic effects of novel antivirals alone or in combination on emergent influenza virus strains warrants further investigation.

A conservation law for virus infection kinetics in vitro.

Conservation laws are among the most important properties of a physical system, but are not commonplace in biology. We derived a conservation law from the basic model for viral infections which consists in a small set of ordinary differential equations. We challenged the conservation law experimentally for the case of a virus infection in a cell culture. We found that the derived, conserved quantity remained almost constant throughout the infection period, implying that the derived conservation law holds in this biological system. We also suggest a potential use for the conservation law in evaluating the accuracy of experimental measurements.

Cost-effectiveness of asenapine in the treatment of bipolar disorder in Canada.

BACKGROUND: Bipolar disorder (BPD) is prevalent and is associated with a significant economic burden. Asenapine, the first tetracyclic antipsychotic approved in Canada for the treatment of BPD, has shown a comparable efficacy profile to other atypical antipsychotics. In addition, it is associated with a favourable metabolic profile and minimal weight gain potential. This study aimed to assess the economic impact of asenapine compared to olanzapine in the treatment of BPD in Canada. METHODS: A decision tree combined with a Markov model was constructed to assess the cost-utility of asenapine compared with olanzapine. The decision tree takes into account the occurrence of extrapyramidal symptoms (EPS), the probability of switching to a different antipsychotic, and the probability of gaining weight. The Markov model takes into account long-term metabolic complications including diabetes, hypertension, coronary heart diseases (CHDs), and stroke. Analyses were conducted from both a Canadian Ministry of Health (MoH) and a societal perspective over a five-year time horizon with yearly cycles. RESULTS: In the treatment of BPD, asenapine is a dominant strategy over olanzapine from both a MoH and a societal perspective. In fact, asenapine is associated with lower costs and more quality-adjusted life years (QALYs). Results of the probabilistic sensitivity analysis indicated that asenapine remains a dominant strategy in 99.2% of the simulations, in both a MoH and a societal perspective, and this result is robust to the many deterministic sensitivity analyses performed. CONCLUSIONS: This economic evaluation demonstrates that asenapine is a cost-effective strategy compared to olanzapine in the treatment of BPD in Canada.

Practical parameter identifiability and handling of censored data with Bayesian inference in mathematical tumour models.

Mechanistic mathematical models (MMs) are a powerful tool to help us understand and predict the dynamics of tumour growth under various conditions. In this work, we use 5 MMs with an increasing number of parameters to explore how certain (often overlooked) decisions in estimating parameters from data of experimental tumour growth affect the outcome of the analysis. In particular, we propose a framework for including tumour volume measurements that fall outside the upper and lower limits of detection, which are normally discarded. We demonstrate how excluding censored data results in an overestimation of the initial tumour volume and the MM-predicted tumour volumes prior to the first measurements, and an underestimation of the carrying capacity and the MM-predicted tumour volumes beyond the latest measurable time points. We show in which way the choice of prior for the MM parameters can impact the posterior distributions, and illustrate that reporting the most likely parameters and their 95% credible interval can lead to confusing or misleading interpretations. We hope this work will encourage others to carefully consider choices made in parameter estimation and to adopt the approaches we put forward herein.

Implementation of Artificial Intelligence-Based Diabetic Retinopathy Screening in a Tertiary Care Hospital in Quebec: Prospective Validation Study.

BACKGROUND: Diabetic retinopathy (DR) affects about 25% of people with diabetes in Canada. Early detection of DR is essential for preventing vision loss. OBJECTIVE: We evaluated the real-world performance of an artificial intelligence (AI) system that analyzes fundus images for DR screening in a Quebec tertiary care center. METHODS: We prospectively recruited adult patients with diabetes at the Centre hospitalier de l'Université de Montréal (CHUM) in Montreal, Quebec, Canada. Patients underwent dual-pathway screening: first by the Computer Assisted Retinal Analysis (CARA) AI system (index test), then by standard ophthalmological examination (reference standard). We measured the AI system's sensitivity and specificity for detecting referable disease at the patient level, along with its performance for detecting any retinopathy and diabetic macular edema (DME) at the eye level, and potential cost savings. RESULTS: This study included 115 patients. CARA demonstrated a sensitivity of 87.5% (95% CI 71.9-95.0) and specificity of 66.2% (95% CI 54.3-76.3) for detecting referable disease at the patient level. For any retinopathy detection at the eye level, CARA showed 88.2% sensitivity (95% CI 76.6-94.5) and 71.4% specificity (95% CI 63.7-78.1). For DME detection, CARA had 100% sensitivity (95% CI 64.6-100) and 81.9% specificity (95% CI 75.6-86.8). Potential yearly savings from implementing CARA at the CHUM were estimated at CAD $245,635 (US $177,643.23, as of July 26, 2024) considering 5000 patients with diabetes. CONCLUSIONS: Our study indicates that integrating a semiautomated AI system for DR screening demonstrates high sensitivity for detecting referable disease in a real-world setting. This system has the potential to improve screening efficiency and reduce costs at the CHUM, but more work is needed to validate it.