Predicting deterioration in dengue using a low cost wearable for continuous clinical monitoring.

Close vital signs monitoring is crucial for the clinical management of patients with dengue. We investigated performance of a non-invasive wearable utilising photoplethysmography (PPG), to provide real-time risk prediction in hospitalised individuals. We performed a prospective observational clinical study in Vietnam between January 2020 and October 2022: 153 patients were included in analyses, providing 1353 h of PPG data. Using a multi-modal transformer approach, 10-min PPG waveform segments and basic clinical data (age, sex, clinical features on admission) were used as features to continuously forecast clinical state 2 h ahead. Prediction of low-risk states (17,939/80,843; 22.1%), defined by NEWS2 and mSOFA < 6, was associated with an area under the precision-recall curve of 0.67 and an area under the receiver operator curve of 0.83. Implementation of such interventions could provide cost-effective triage and clinical care in dengue, offering opportunities for safe ambulatory patient management.

Recent Electrochemical Advancements for Liquid-Biopsy Nucleic Acid Detection for Point-of-Care Prostate Cancer Diagnostics and Prognostics.

Current diagnostic and prognostic tests for prostate cancer require specialised laboratories and have low specificity for prostate cancer detection. As such, recent advancements in electrochemical devices for point of care (PoC) prostate cancer detection have seen significant interest. Liquid-biopsy detection of relevant circulating and exosomal nucleic acid markers presents the potential for minimally invasive testing. In combination, electrochemical devices and circulating DNA and RNA detection present an innovative approach for novel prostate cancer diagnostics, potentially directly within the clinic. Recent research in electrochemical impedance spectroscopy, voltammetry, chronoamperometry and potentiometric sensing using field-effect transistors will be discussed. Evaluation of the PoC relevance of these techniques and their fulfilment of the WHO's REASSURED criteria for medical diagnostics is described. Further areas for exploration within electrochemical PoC testing and progression to clinical implementation for prostate cancer are assessed.

An automatic deep reinforcement learning bolus calculator for automated insulin delivery systems.

In hybrid automatic insulin delivery (HAID) systems, meal disturbance is compensated by feedforward control, which requires the announcement of the meal by the patient with type 1 diabetes (DM1) to achieve the desired glycemic control performance. The calculation of insulin bolus in the HAID system is based on the amount of carbohydrates (CHO) in the meal and patient-specific parameters, i.e. carbohydrate-to-insulin ratio (CR) and insulin sensitivity-related correction factor (CF). The estimation of CHO in a meal is prone to errors and is burdensome for patients. This study proposes a fully automatic insulin delivery (FAID) system that eliminates patient intervention by compensating for unannounced meals. This study exploits the deep reinforcement learning (DRL) algorithm to calculate insulin bolus for unannounced meals without utilizing the information on CHO content. The DRL bolus calculator is integrated with a closed-loop controller and a meal detector (both previously developed by our group) to implement the FAID system. An adult cohort of 68 virtual patients based on the modified UVa/Padova simulator was used for in-silico trials. The percentage of the overall duration spent in the target range of 70-180 mg/dL was 71.2 % and 76.2 % , < 70 mg/dL was 0.9 % and 0.1 % , and > 180 mg/dL was 26.7 % and 21.1 % , respectively, for the FAID system and HAID system utilizing a standard bolus calculator (SBC) including CHO misestimation. The proposed algorithm can be exploited to realize FAID systems in the future.

Multi-Horizon Glucose Prediction Across Populations with Deep Domain Generalization.

Real-time continuous glucose monitoring (CGM), augmented with accurate glucose prediction, offers an effective strategy for maintaining blood glucose levels within a therapeutically appropriate range. This is particularly crucial for individuals with type 1 diabetes (T1D) who require long-term self-management. However, with extensive glycemic variability, developing a prediction algorithm applicable across diverse populations remains a significant challenge. Leveraging meta-learning for domain generalization, we propose GPFormer, a Transformer-based zero-shot learning method designed for multi-horizon glucose prediction. We developed GPFormer on the REPLACE-BG dataset, comprising 226 participants with T1D, and proceeded to evaluate its performance using three external clinical datasets with CGM data. These included the OhioT1DM dataset, a publicly available dataset including 12 T1D participants, as well as two proprietary datasets. The first proprietary dataset included 22 participants, while the second contained 45 participants, encompassing a diverse group with T1D, type 2 diabetes, and those without diabetes, including patients admitted to hospitals. These four datasets include both outpatient and inpatient settings, various intervention strategies, and demographic variability, which effectively reflect real-world scenarios of CGM usage. When compared with a group of machine learning baseline methods, GPFormer consistently demonstrated superior performance and achieved the lowest root mean square error for all the evaluated datasets up to a prediction horizon of two hours. These experimental results highlight the effectiveness and generalizability of the proposed model across a variety of populations, demonstrating its substantial potential to enhance glucose management in a wide range of practical clinical settings.

Rapid, Portable, and Electricity-free Sample Extraction Method for Enhanced Molecular Diagnostics in Resource-Limited Settings.

The COVID-19 pandemic has highlighted the need for rapid and reliable diagnostics that are accessible in resource-limited settings. To address this pressing issue, we have developed a rapid, portable, and electricity-free method for extracting nucleic acids from respiratory swabs (i.e. nasal, nasopharyngeal and buccal swabs), successfully demonstrating its effectiveness for the detection of SARS-CoV-2 in residual clinical specimens. Unlike traditional approaches, our solution eliminates the need for micropipettes or electrical equipment, making it user-friendly and requiring little to no training. Our method builds upon the principles of magnetic bead extraction and revolves around a low-cost plastic magnetic lid, called SmartLid, in combination with a simple disposable kit containing all required reagents conveniently prealiquoted. Here, we clinically validated the SmartLid sample preparation method in comparison to the gold standard QIAamp Viral RNA Mini Kit from QIAGEN, using 406 clinical isolates, including 161 SARS-CoV-2 positives, using the SARS-CoV-2 RT-qPCR assays developed by the US Centers for Disease Control and Prevention (CDC). The SmartLid method showed an overall sensitivity of 95.03% (95% CI: 90.44-97.83%) and a specificity of 99.59% (95% CI: 97.76-99.99%), with a positive agreement of 97.79% (95% CI: 95.84-98.98%) when compared to QIAGEN's column-based extraction method. There are clear benefits to using the SmartLid sample preparation kit: it enables swift extraction of viral nucleic acids, taking less than 5 min, without sacrificing significant accuracy when compared to more expensive and time-consuming alternatives currently available on the market. Moreover, its simplicity makes it particularly well-suited for the point-of-care where rapid results and portability are crucial. By providing an efficient and accessible means of nucleic acid extraction, our approach aims to introduce a step-change in diagnostic capabilities for resource-limited settings.

Biocompatibility characterisation of CMOS-based Lab-on-Chip electrochemical sensors for in vitro cancer cell culture applications.

Lab-on-Chip electrochemical sensors, such as Ion-Sensitive Field-Effect Transistors (ISFETs), are being developed for use in point-of-care diagnostics, such as pH detection of tumour microenvironments, due to their integration with standard Complementary Metal Oxide Semiconductor (CMOS) technology. With this approach, the passivation of the CMOS process is used as a sensing layer to minimise post-processing, and Silicon Nitride (Si3N4) is the most common material at the microchip surface. ISFETs have the potential to be used for cell-based assays however, there is a poor understanding of the biocompatibility of microchip surfaces. Here, we quantitatively evaluated cell adhesion, morphogenesis, proliferation and mechano-responsiveness of both normal and cancer cells cultured on a Si3N4, sensor surface. We demonstrate that both normal and cancer cell adhesion decreased on Si3N4. Activation of the mechano-responsive transcription regulators, YAP/TAZ, are significantly decreased in cancer cells on Si3N4 in comparison to standard cell culture plastic, whilst proliferation marker, Ki67, expression markedly increased. Non-tumorigenic cells on chip showed less sensitivity to culture on Si3N4 than cancer cells. Treatment with extracellular matrix components increased cell adhesion in normal and cancer cell cultures, surpassing the adhesiveness of plastic alone. Moreover, poly-l-ornithine and laminin treatment restored YAP/TAZ levels in both non-tumorigenic and cancer cells to levels comparable to those observed on plastic. Thus, engineering the electrochemical sensor surface with treatments will provide a more physiologically relevant environment for future cell-based assay development on chip.

Towards a machine-learning assisted non-invasive classification of dengue severity using wearable PPG data: a prospective clinical study.

BACKGROUND: Dengue epidemics impose considerable strain on healthcare resources. Real-time continuous and non-invasive monitoring of patients admitted to the hospital could lead to improved care and outcomes. We evaluated the performance of a commercially available wearable (SmartCare) utilising photoplethysmography (PPG) to stratify clinical risk for a cohort of hospitalised patients with dengue in Vietnam. METHODS: We performed a prospective observational study for adult and paediatric patients with a clinical diagnosis of dengue at the Hospital for Tropical Disease, Ho Chi Minh City, Vietnam. Patients underwent PPG monitoring early during admission alongside standard clinical care. PPG waveforms were analysed using machine learning models. Adult patients were classified between 3 severity classes: i) uncomplicated (ward-based), ii) moderate-severe (emergency department-based), and iii) severe (ICU-based). Data from paediatric patients were split into 2 classes: i) severe (during ICU stay) and ii) follow-up (14-21 days after the illness onset). Model performances were evaluated using standard classification metrics and 5-fold stratified cross-validation. FINDINGS: We included PPG and clinical data from 132 adults and 15 paediatric patients with a median age of 28 (IQR, 21-35) and 12 (IQR, 9-13) years respectively. 1781 h of PPG data were available for analysis. The best performing convolutional neural network models (CNN) achieved a precision of 0.785 and recall of 0.771 in classifying adult patients according to severity class and a precision of 0.891 and recall of 0.891 in classifying between disease and post-disease state in paediatric patients. INTERPRETATION: We demonstrate that the use of a low-cost wearable provided clinically actionable data to differentiate between patients with dengue of varying severity. Continuous monitoring and connectivity to early warning systems could significantly benefit clinical care in dengue, particularly within an endemic setting. Work is currently underway to implement these models for dynamic risk predictions and assist in individualised patient care. FUNDING: EPSRC Centre for Doctoral Training in High-Performance Embedded and Distributed Systems (HiPEDS) (Grant: EP/L016796/1) and the Wellcome Trust (Grants: 215010/Z/18/Z and 215688/Z/19/Z).

Personalising intravenous to oral antibiotic switch decision making through fair interpretable machine learning.

Antimicrobial resistance (AMR) and healthcare associated infections pose a significant threat globally. One key prevention strategy is to follow antimicrobial stewardship practices, in particular, to maximise targeted oral therapy and reduce the use of indwelling vascular devices for intravenous (IV) administration. Appreciating when an individual patient can switch from IV to oral antibiotic treatment is often non-trivial and not standardised. To tackle this problem we created a machine learning model to predict when a patient could switch based on routinely collected clinical parameters. 10,362 unique intensive care unit stays were extracted and two informative feature sets identified. Our best model achieved a mean AUROC of 0.80 (SD 0.01) on the hold-out set while not being biased to individuals protected characteristics. Interpretability methodologies were employed to create clinically useful visual explanations. In summary, our model provides individualised, fair, and interpretable predictions for when a patient could switch from IV-to-oral antibiotic treatment. Prospectively evaluation of safety and efficacy is needed before such technology can be applied clinically.

Population-Specific Glucose Prediction in Diabetes Care With Transformer-Based Deep Learning on the Edge.

Leveraging continuous glucose monitoring (CGM) systems, real-time blood glucose (BG) forecasting is essential for proactive interventions, playing a crucial role in enhancing the management of type 1 diabetes (T1D) and type 2 diabetes (T2D). However, developing a model generalized to a population and subsequently embedding it within a microchip of a wearable device presents significant technical challenges. Furthermore, the domain of BG prediction in T2D remains under-explored in the literature. In light of this, we propose a population-specific BG prediction model, leveraging the capabilities of the temporal fusion Transformer (TFT) to adjust predictions based on personal demographic data. Then the trained model is embedded within a system-on-chip, integral to our low-power and low-cost customized wearable device. This device seamlessly communicates with CGM systems through Bluetooth and provides timely BG predictions using edge computing. When evaluated on two publicly available clinical datasets with a total of 124 participants with T1D or T2D, the embedded TFT model consistently demonstrated superior performance, achieving the lowest prediction errors when compared with a range of machine learning baseline methods. Executing the TFT model on our wearable device requires minimal memory and power consumption, enabling continuous decision support for more than 51 days on a single Li-Poly battery charge. These findings demonstrate the significant potential of the proposed TFT model and wearable device in enhancing the quality of life for people with diabetes and effectively addressing real-world challenges.

The effect of nodal connectivity and strut density within stochastic titanium scaffolds on osteogenesis.

Modern orthopaedic implants use lattice structures that act as 3D scaffolds to enhance bone growth into and around implants. Stochastic scaffolds are of particular interest as they mimic the architecture of trabecular bone and can combine isotropic properties and adjustable structure. The existing research mainly concentrates on controlling the mechanical and biological performance of periodic lattices by adjusting pore size and shape. Still, less is known on how we can control the performance of stochastic lattices through their design parameters: nodal connectivity, strut density and strut thickness. To elucidate this, four lattice structures were evaluated with varied strut densities and connectivity, hence different local geometry and mechanical properties: low apparent modulus, high apparent modulus, and two with near-identical modulus. Pre-osteoblast murine cells were seeded on scaffolds and cultured in vitro for 28 days. Cell adhesion, proliferation and differentiation were evaluated. Additionally, the expression levels of key osteogenic biomarkers were used to assess the effect of each design parameter on the quality of newly formed tissue. The main finding was that increasing connectivity increased the rate of osteoblast maturation, tissue formation and mineralisation. In detail, doubling the connectivity, over fixed strut density, increased collagen type-I by 140%, increased osteopontin by 130% and osteocalcin by 110%. This was attributed to the increased number of acute angles formed by the numerous connected struts, which facilitated the organization of cells and accelerated the cell cycle. Overall, increasing connectivity and adjusting strut density is a novel technique to design stochastic structures which combine a broad range of biomimetic properties and rapid ossification.

Smart-Plexer: a breakthrough workflow for hybrid development of multiplex PCR assays.

Developing multiplex PCR assays requires extensive experimental testing, the number of which exponentially increases by the number of multiplexed targets. Dedicated efforts must be devoted to the design of optimal multiplex assays ensuring specific and sensitive identification of multiple analytes in a single well reaction. Inspired by data-driven approaches, we reinvent the process of developing and designing multiplex assays using a hybrid, simple workflow, named Smart-Plexer, which couples empirical testing of singleplex assays and computer simulation to develop optimised multiplex combinations. The Smart-Plexer analyses kinetic inter-target distances between amplification curves to generate optimal multiplex PCR primer sets for accurate multi-pathogen identification. In this study, the Smart-Plexer method is applied and evaluated for seven respiratory infection target detection using an optimised multiplexed PCR assay. Single-channel multiplex assays, together with the recently published data-driven methodology, Amplification Curve Analysis (ACA), were demonstrated to be capable of classifying the presence of desired targets in a single test for seven common respiratory infection pathogens.

Handheld ISFET Lab-on-Chip detection of TMPRSS2-ERG and AR mRNA for prostate cancer prognostics.

Ion-sensitive field-effect transistors (ISFETs) in combination with unmodified complementary metal oxide semiconductors present a point-of-care platform for clinical diagnostics and prognostics. This work illustrates the sensitive and specific detection of two circulating mRNA markers for prostate cancer, the androgen receptor and the TMPRSS2-ERG fusion using a target-specific loop-mediated isothermal amplification method. TMPRSS2-ERG and androgen receptor RNA were detected down to 3x101 and 5x101 copies respectively in under 30 minutes. Administration of these assays onto the ISFET Lab-on-chip device was successful and the specificity of each marker was corroborated with mRNA extracted from prostate cancer cell lines.

Analytical and diagnostic performance characteristics of reverse-transcriptase loop-mediated isothermal amplification assays for dengue virus serotypes 1-4: A scoping review to inform potential use in portable molecular diagnostic devices.

Dengue is a mosquito-borne disease caused by dengue virus (DENV) serotypes 1-4 which affects 100-400 million adults and children each year. Reverse-transcriptase (RT) quantitative polymerase chain reaction (qPCR) assays are the current gold-standard in diagnosis and serotyping of infections, but their use in low-middle income countries (LMICs) has been limited by laboratory infrastructure requirements. Loop-mediated isothermal amplification (LAMP) assays do not require thermocycling equipment and therefore could potentially be deployed outside laboratories and/or miniaturised. This scoping literature review aimed to describe the analytical and diagnostic performance characteristics of previously developed serotype-specific dengue RT-LAMP assays and evaluate potential for use in portable molecular diagnostic devices. A literature search in Medline was conducted. Studies were included if they were listed before 4th May 2022 (no prior time limit set) and described the development of any serotype-specific DENV RT-LAMP assay ('original assays') or described the further evaluation, adaption or implementation of these assays. Technical features, analytical and diagnostic performance characteristics were collected for each assay. Eight original assays were identified. These were heterogenous in design and reporting. Assays' lower limit of detection (LLOD) and linear range of quantification were comparable to RT-qPCR (with lowest reported values 2.2x101 and 1.98x102 copies/ml, respectively, for studies which quantified target RNA copies) and analytical specificity was high. When evaluated, diagnostic performance was also high, though reference diagnostic criteria varied widely, prohibiting comparison between assays. Fourteen studies using previously described assays were identified, including those where reagents were lyophilised or 'printed' into microfluidic channels and where several novel detection methods were used. Serotype-specific DENV RT-LAMP assays are high-performing and have potential to be used in portable molecular diagnostic devices if they can be integrated with sample extraction and detection methods. Standardised reporting of assay validation and diagnostic accuracy studies would be beneficial.

Offline Deep Reinforcement Learning and Off-Policy Evaluation for Personalized Basal Insulin Control in Type 1 Diabetes.

Recent advancements in hybrid closed-loop systems, also known as the artificial pancreas (AP), have been shown to optimize glucose control and reduce the self-management burdens for people living with type 1 diabetes (T1D). AP systems can adjust the basal infusion rates of insulin pumps, facilitated by real-time communication with continuous glucose monitoring. Deep reinforcement learning (DRL) has introduced new paradigms of basal insulin control algorithms. However, all the existing DRL-based AP controllers require extensive random online interactions between the agent and environment. While this can be validated in T1D simulators, it becomes impractical in real-world clinical settings. To this end, we propose an offline DRL framework that can develop and validate models for basal insulin control entirely offline. It comprises a DRL model based on the twin delayed deep deterministic policy gradient and behavior cloning, as well as off-policy evaluation (OPE) using fitted Q evaluation. We evaluated the proposed framework on an in silico dataset generated by the UVA/Padova T1D simulator, and the OhioT1DM dataset, a real clinical dataset. The performance on the in silico dataset shows that the offline DRL algorithm significantly increased time in range while reducing time below range and time above range for both adult and adolescent groups. Then, we used the OPE to estimate model performance on the clinical dataset, where a notable increase in policy values was observed for each subject. The results demonstrate that the proposed framework is a viable and safe method for improving personalized basal insulin control in T1D.

Development and Validation of Binary Classifiers to Predict Nocturnal Hypoglycemia in Adults With Type 1 Diabetes.

BACKGROUND: One of the biggest challenges for people with type 1 diabetes (T1D) using multiple daily injections (MDIs) is nocturnal hypoglycemia (NH). Recurrent NH can lead to serious complications; hence, prevention is of high importance. In this work, we develop and externally validate, device-agnostic Machine Learning (ML) models to provide bedtime decision support to people with T1D and minimize the risk of NH. METHODS: We present the design and development of binary classifiers to predict NH (blood glucose levels occurring below 70 mg/dL). Using data collected from a 6-month study of 37 adult participants with T1D under free-living conditions, we extract daytime features from continuous glucose monitor (CGM) sensors, administered insulin, meal, and physical activity information. We use these features to train and test the performance of two ML algorithms: Random Forests (RF) and Support Vector Machines (SVMs). We further evaluate our model in an external population of 20 adults with T1D using MDI insulin therapy and wearing CGM and flash glucose monitoring sensors for two periods of eight weeks each. RESULTS: At population-level, SVM outperforms RF algorithm with a receiver operating characteristic-area under curve (ROC-AUC) of 79.36% (95% CI: 76.86%, 81.86%). The proposed SVM model generalizes well in an unseen population (ROC-AUC = 77.06%), as well as between the two different glucose sensors (ROC-AUC = 77.74%). CONCLUSIONS: Our model shows state-of-the-art performance, generalizability, and robustness in sensor devices from different manufacturers. We believe it is a potential viable approach to inform people with T1D about their risk of NH before it occurs.

Improved Stability of Graphene-Coated CMOS ISFETs for Biosensing.

A polymer-assisted graphene transfer method is used to transfer sheets of monolayer and multilayer graphene onto the passivation layer of ion-sensitive field effect transistor arrays. The arrays are fabricated using commercial 0.35 µm complementary metal-oxide-semiconductor (CMOS) technology and contain 3874 pixels sensitive to pH changes on the top silicon nitride surface. By inhibiting dispersive ion transport and hydration of this underlying nitride layer, the transferred graphene sheets help address non-idealities in the sensor response while retaining some pH sensitivity due to the presence of ion adsorption sites. Improvements in hydrophilicity and electrical conductivity of the sensing surface after graphene transfer, as well as in-plane molecular diffusion along the graphene-nitride interface, also greatly improve spatial consistency across an array, allowing for ∼20% more pixels to remain within operating range and enhancing sensor reliability. Multilayer graphene offers a better performance trade-off than monolayer graphene, reducing drift rate by ∼25% and drift amplitude by ∼59% with minimal reduction in pH sensitivity. Monolayer graphene offers slightly better temporal and spatial uniformity in performance of a sensing array, which is associated with the consistency in layer thickness and a lower defect density.

Applying artificial intelligence and digital health technologies, Viet Nam.

Problem: Direct application of digital health technologies from high-income settings to low- and middle-income countries may be inappropriate due to challenges around data availability, implementation and regulation. Hence different approaches are needed. Approach: Within the Viet Nam ICU Translational Applications Laboratory project, since 2018 we have been developing a wearable device for individual patient monitoring and a clinical assessment tool to improve dengue disease management. Working closely with local staff at the Hospital for Tropical Diseases, Ho Chi Minh City, we developed and tested a prototype of the wearable device. We obtained perspectives on design and use of the sensor from patients. To develop the assessment tool, we used existing research data sets, mapped workflows and clinical priorities, interviewed stakeholders and held workshops with hospital staff. Local setting: In Viet Nam, a lower middle-income country, the health-care system is in the nascent stage of implementing digital health technologies. Relevant changes: Based on patient feedback, we are altering the design of the wearable sensor to increase comfort. We built the user interface of the assessment tool based on the core functionalities selected by workshop attendees. The interface was subsequently tested for usability in an iterative manner by the clinical staff members. Lessons learnt: The development and implementation of digital health technologies need an interoperable and appropriate plan for data management including collection, sharing and integration. Engagements and implementation studies should be conceptualized and conducted alongside the digital health technology development. The priorities of end-users, and understanding context and regulatory landscape are crucial for success.

A dual paper-based nucleic acid extraction method from blood in under ten minutes for point-of-care diagnostics.

Nucleic acid extraction (NAE) plays a crucial role for diagnostic testing procedures. For decades, dried blood spots (DBS) have been used for serology, drug monitoring, and molecular studies. However, extracting nucleic acids from DBS remains a significant challenge, especially when attempting to implement these applications to the point-of-care (POC). To address this issue, we have developed a paper-based NAE method using cellulose filter papers (DBSFP) that operates without the need for electricity (at room temperature). Our method allows for NAE in less than 7 min, and it involves grade 3 filter paper pre-treated with 8% (v/v) igepal surfactant, 1 min washing step with 1× PBS, and 5 min incubation at room temperature in 1× TE buffer. The performance of the methodology was assessed with loop-mediated isothermal amplification (LAMP), targeting the human reference gene beta-actin and the kelch 13 gene from P. falciparum. The developed method was evaluated against FTA cards and magnetic bead-based purification, using time-to-positive (min) for comparative analysis. Furthermore, we optimised our approach to take advantage of the dual functionality of the paper-based extraction, allowing for elution (eluted disk) as well as direct placement of the disk in the LAMP reaction (in situ disk). This flexibility extends to eukaryotic cells, bacterial cells, and viral particles. We successfully validated the method for RNA/DNA detection and demonstrated its compatibility with whole blood stored in anticoagulants. Additionally, we studied the compatibility of DBSFP with colorimetric and lateral flow detection, showcasing its potential for POC applications. Across various tested matrices, targets, and experimental conditions, our results were comparable to those obtained using gold standard methods, highlighting the versatility of our methodology. In summary, this manuscript presents a cost-effective solution for NAE from DBS, enabling molecular testing in virtually any POC setting. When combined with LAMP, our approach provides sample-to-result detection in under 35 minutes.

GluGAN: Generating Personalized Glucose Time Series Using Generative Adversarial Networks.

Time series data generated by continuous glucose monitoring sensors offer unparalleled opportunities for developing data-driven approaches, especially deep learning-based models, in diabetes management. Although these approaches have achieved state-of-the-art performance in various fields such as glucose prediction in type 1 diabetes (T1D), challenges remain in the acquisition of large-scale individual data for personalized modeling due to the elevated cost of clinical trials and data privacy regulations. In this work, we introduce GluGAN, a framework specifically designed for generating personalized glucose time series based on generative adversarial networks (GANs). Employing recurrent neural network (RNN) modules, the proposed framework uses a combination of unsupervised and supervised training to learn temporal dynamics in latent spaces. Aiming to assess the quality of synthetic data, we apply clinical metrics, distance scores, and discriminative and predictive scores computed by post-hoc RNNs in evaluation. Across three clinical datasets with 47 T1D subjects (including one publicly available and two proprietary datasets), GluGAN achieved better performance for all the considered metrics when compared with four baseline GAN models. The performance of data augmentation is evaluated by three machine learning-based glucose predictors. Using the training sets augmented by GluGAN significantly reduced the root mean square error for the predictors over 30 and 60-minute horizons. The results suggest that GluGAN is an effective method in generating high-quality synthetic glucose time series and has the potential to be used for evaluating the effectiveness of automated insulin delivery algorithms and as a digital twin to substitute for pre-clinical trials.

Sensitive Detection of Asymptomatic and Symptomatic Malaria with Seven Novel Parasite-Specific LAMP Assays and Translation for Use at Point-of-Care.

Human malaria is a life-threatening parasitic disease with high impact in the sub-Saharan Africa region, where 95% of global cases occurred in 2021. While most malaria diagnostic tools are focused on Plasmodium falciparum, there is a current lack of testing non-P. falciparum cases, which may be underreported and, if undiagnosed or untreated, may lead to severe consequences. In this work, seven species-specific loop-mediated isothermal amplification (LAMP) assays were designed and evaluated against TaqMan quantitative PCR (qPCR), microscopy, and enzyme-linked immunosorbent assays (ELISAs). Their clinical performance was assessed with a cohort of 164 samples of symptomatic and asymptomatic patients from Ghana. All asymptomatic samples with a parasite load above 80 genomic DNA (gDNA) copies per µL of extracted sample were detected with the Plasmodium falciparum LAMP assay, reporting 95.6% (95% confidence interval [95% CI] of 89.9 to 98.5) sensitivity and 100% (95% CI of 87.2 to 100) specificity. This assay showed higher sensitivity than microscopy and ELISA, which were 52.7% (95% CI of 39.7 to 67%) and 67.3% (95% CI of 53.3 to 79.3%), respectively. Nine samples were positive for P. malariae, indicating coinfections with P. falciparum, which represented 5.5% of the tested population. No samples were detected as positive for P. vivax, P. ovale, P. knowlesi, or P. cynomolgi by any method. Furthermore, translation to the point-of-care was demonstrated with a subcohort of 18 samples tested locally in Ghana using our handheld lab-on-chip platform, Lacewing, showing comparable results to a conventional fluorescence-based instrument. The developed molecular diagnostic test could detect asymptomatic malaria cases, including submicroscopic parasitemia, and it has the potential to be used for point-of-care applications. IMPORTANCE The spread of Plasmodium falciparum parasites with Pfhrp2/3 gene deletions presents a major threat to reliable point-of-care diagnosis with current rapid diagnostic tests (RDTs). Novel molecular diagnostics based on nucleic acid amplification are needed to address this liability. In this work, we overcome this challenge by developing sensitive tools for the detection of Plasmodium falciparum and non-P. falciparum species. Furthermore, we evaluate these tools with a cohort of symptomatic and asymptomatic malaria patients and test a subcohort locally in Ghana. The findings of this work could lead to the implementation of DNA-based diagnostics to fight against the spread of malaria and provide reliable, sensitive, and specific diagnostics at the point of care.