Experimental pneumococcal carriage in people living with HIV in Malawi: the first controlled human infection model in a key at-risk population.

Background: As well as suffering a high burden of pneumococcal disease people living with HIV (PLHIV) may contribute to community transmission in sub-Saharan African (sSA) settings. Pneumococcal vaccination is not currently offered to PLHIV in sSA but may prevent disease and reduce transmission. More evidence of vaccine effectiveness against carriage in PLHIV is needed. An Experimental Human Pneumococcal Carriage model (EHPC) has been safely and acceptably used in healthy adults in Malawi to evaluate pneumococcal vaccines against carriage and to identify immune correlates of protection from carriage. This study will establish the same model in PLHIV and will be the first controlled human infection model (CHIM) in this key population. Methods: Healthy participants with and without HIV will be inoculated intranasally with Streptococcus pneumoniae serotype 6B. Sequential cohorts will be challenged with increasing doses to determine the optimal safe challenge dose to establish experimental carriage. Nasal fluid, nasal mucosal, and blood samples will be taken before inoculation and on days 2, 7, 14, and 21 following inoculation to measure pneumococcal carriage density and identify immune correlates of protection from carriage. The vast majority of natural pneumococcal carriage events in PLHIV do not result in invasive disease and no invasive disease is expected in this study. However, robust participant safety monitoring is designed to identify signs of invasive disease early should they develop, and to implement treatment immediately. Participants will complete a Likert-style questionnaire at study-end to establish acceptability. Interpretations: We expect the EHPC model to be safely and acceptably implemented in PLHIV. The CHIM can then be used to accelerate pneumococcal vaccine evaluations in this population, and an evidence-based pneumococcal vaccination policy for PLHIV in sSA.

Prevalence, patterns and associated risk factors for dyslipidaemia among individuals attending the diabetes clinic at a tertiary hospital in Central Malawi.

BACKGROUND: Dyslipidaemia among individuals with diabetes is a significant modifiable risk factor for atherosclerotic cardiovascular diseases (ASCVDs). ASCVDs are a major cause of mortality and morbidity globally, especially in people with diabetes. In Malawi, limited data exist on the prevalence and biochemical characteristics of diabetic dyslipidaemia. This study investigated the prevalence and biochemical characteristics of dyslipidaemia in individuals attending the diabetes clinic at Kamuzu Central Hospital, the largest tertiary referral hospital in Central Malawi. METHODS: Using a cross-sectional design, sociodemographic, medical and anthropometric data were collected from 391 adult participants who were enrolled in the study. Blood samples were analysed for glycosylated haemoglobin (HBA1c) and fasting lipid profiles. The prevalence of dyslipidaemia was calculated, and the biochemical characteristics of the dyslipidaemia were defined. The associations between dyslipidaemia and risk factors such as sociodemographic characteristics, obesity, and HBA1c levels were evaluated using logistic regression analysis. RESULTS: Prevalence of dyslipidaemia was observed in 71% of the participants, and elevated low-density lipoprotein cholesterol was the most frequent lipid abnormality among the study participants. None of the participants were receiving any lipid-lowering therapy. On bivariate analysis, dyslipidemia was positively associated with female sex [OR 1.65 (95% CI 1.05- 2.58); p = 0.09], age ≥ 30 years [OR 3.60 (95% CI 1.17-7.68); p = 0.001] and overweight and obesity [OR 2.11 (95% CI 1.33-3.34); p = 0.002]. On multivariate analysis, being overweight or obese was an independent predictor of dyslipidaemia [AOR 1.8;(95% CI 1.15- 3.37); p = 0.04]. CONCLUSION: Dyslipidaemia was highly prevalent among individuals with diabetes in this study, and elevated low-density lipoprotein cholesterol was the most frequent lipid abnormality. Overweight and obesity were also highly prevalent and positively predicted dyslipidaemia. This study highlights the importance of appropriately addressing dyslipidaemia, overweight and obesity among individuals with diabetes in Malawi and other similar settings in Africa as one of the significant ways of reducing the risk of ASCVDs among this population.

Prevalence, patterns and associated risk factors for dyslipidaemia among individuals attending the diabetes clinic at a tertiary hospital in Central Malawi.

Background: Dyslipidaemia among individuals with diabetes is a significant modifiable risk factor for atherosclerotic cardiovascular diseases (ASCVDs). ASCVDs are a major cause of mortality and morbidity globally, especially in people with diabetes. In Malawi, limited data exist on the prevalence and biochemical characteristics of diabetic dyslipidaemia. This study investigated the prevalence and biochemical characteristics of dyslipidaemia in individuals attending the diabetes clinic at Kamuzu Central Hospital, the largest tertiary referral hospital in Central Malawi. Methods: Using a cross-sectional design, sociodemographic, medical and anthropometric data were collected from 391 adult participants who were enrolled in the study. Blood samples were analysed for glycosylated haemoglobin (HBA1c) and fasting lipid profiles. The prevalence of dyslipidaemia was calculated, and the biochemical characteristics of the dyslipidaemia were defined. The associations between dyslipidaemia and risk factors such as sociodemographic characteristics, obesity, and HBA1c levels were evaluated using logistic regression analysis. Results: Prevalence of dyslipidaemia was observed in 71% of the participants, and elevated low-density lipoprotein cholesterol was the most frequent lipid abnormality among the study participants. On bivariate analysis, dyslipidemia was positively associated with female sex [OR 1.65 (95% CI 1.05-2.58); p = 0.09], age ≥ 30 years [OR 3.60 (95% CI 1.17-7.68); p = 0.001] and overweight and obesity [OR 2.11 (95% CI 1.33-3.34); p = 0.002]. On multivariate analysis, being overweight or obese was an independent predictor of dyslipidaemia [AOR 1.8 ;( 95% CI 1.15-3.37); p = 0.04]. Conclusion: Dyslipidaemia was highly prevalent among individuals with diabetes in this study, and elevated low-density lipoprotein cholesterol was the most frequent lipid abnormality. Overweight and obesity were also highly prevalent and positively predicted dyslipidaemia. This study highlights the importance of appropriately addressing dyslipidaemia, overweight and obesity among individuals with diabetes in Malawi and other similar settings in Africa as one of the significant ways of reducing the risk of ASCVDs among this population.

Crowdsourcing digital health measures to predict Parkinson's disease severity: the Parkinson's Disease Digital Biomarker DREAM Challenge.

Consumer wearables and sensors are a rich source of data about patients' daily disease and symptom burden, particularly in the case of movement disorders like Parkinson's disease (PD). However, interpreting these complex data into so-called digital biomarkers requires complicated analytical approaches, and validating these biomarkers requires sufficient data and unbiased evaluation methods. Here we describe the use of crowdsourcing to specifically evaluate and benchmark features derived from accelerometer and gyroscope data in two different datasets to predict the presence of PD and severity of three PD symptoms: tremor, dyskinesia, and bradykinesia. Forty teams from around the world submitted features, and achieved drastically improved predictive performance for PD status (best AUROC = 0.87), as well as tremor- (best AUPR = 0.75), dyskinesia- (best AUPR = 0.48) and bradykinesia-severity (best AUPR = 0.95).

Gene selection for optimal prediction of cell position in tissues from single-cell transcriptomics data.

Single-cell RNA-sequencing (scRNAseq) technologies are rapidly evolving. Although very informative, in standard scRNAseq experiments, the spatial organization of the cells in the tissue of origin is lost. Conversely, spatial RNA-seq technologies designed to maintain cell localization have limited throughput and gene coverage. Mapping scRNAseq to genes with spatial information increases coverage while providing spatial location. However, methods to perform such mapping have not yet been benchmarked. To fill this gap, we organized the DREAM Single-Cell Transcriptomics challenge focused on the spatial reconstruction of cells from the Drosophila embryo from scRNAseq data, leveraging as silver standard, genes with in situ hybridization data from the Berkeley Drosophila Transcription Network Project reference atlas. The 34 participating teams used diverse algorithms for gene selection and location prediction, while being able to correctly localize clusters of cells. Selection of predictor genes was essential for this task. Predictor genes showed a relatively high expression entropy, high spatial clustering and included prominent developmental genes such as gap and pair-rule genes and tissue markers. Application of the top 10 methods to a zebra fish embryo dataset yielded similar performance and statistical properties of the selected genes than in the Drosophila data. This suggests that methods developed in this challenge are able to extract generalizable properties of genes that are useful to accurately reconstruct the spatial arrangement of cells in tissues.

Excess of singleton loss-of-function variants in Parkinson's disease contributes to genetic risk.

BACKGROUND: Parkinson's disease (PD) is a neurodegenerative disorder with complex genetic architecture. Besides rare mutations in high-risk genes related to monogenic familial forms of PD, multiple variants associated with sporadic PD were discovered via association studies. METHODS: We studied the whole-exome sequencing data of 340 PD cases and 146 ethnically matched controls from the Parkinson's Progression Markers Initiative (PPMI) and performed burden analysis for different rare variant classes. Disease prediction models were built based on clinical, non-clinical and genetic features, including both common and rare variants, and two machine learning methods. RESULTS: We observed a significant exome-wide burden of singleton loss-of-function variants (corrected p=0.037). Overall, no exome-wide burden of rare amino acid changing variants was detected. Finally, we built a disease prediction model combining singleton loss-of-function variants, a polygenic risk score based on common variants, and family history of PD as features and reached an area under the curve of 0.703 (95% CI 0.698 to 0.708). By incorporating a rare variant feature, our model increased the performance of the state-of-the-art classification model for the PPMI dataset, which reached an area under the curve of 0.639 based on common variants alone. CONCLUSION: The main finding of this study is to highlight the contribution of singleton loss-of-function variants to the complex genetics of PD and that disease risk prediction models combining singleton and common variants can improve models built solely on common variants.

Patient outcomes associated with post-tuberculosis lung damage in Malawi: a prospective cohort study.

BACKGROUND: Post-tuberculosis lung damage (PTLD) is a recognised consequence of pulmonary TB (pTB). However, little is known about its prevalence, patterns and associated outcomes, especially in sub-Saharan Africa and HIV-positive adults. METHODS: Adult (≥15 years) survivors of a first episode of pTB in Blantyre, Malawi, completed the St George's Respiratory Questionnaire, 6-minute walk test, spirometry and high-resolution CT (HRCT) chest imaging at TB treatment completion. Symptom, spirometry, health seeking, TB-retreatment and mortality data were collected prospectively to 1 year. Risk factors for persistent symptoms, pulmonary function decline and respiratory-related health-seeking were identified through multivariable regression modelling. RESULTS: Between February 2016 and April 2017, 405 participants were recruited. Median age was 35 years (IQR: 28 to 41), 77.3% (313/405) had had microbiologically proven pTB, and 60.3% (244/403) were HIV-positive. At pTB treatment completion, 60.7% (246/405) reported respiratory symptoms, 34.2% (125/365) had abnormal spirometry, 44.2% (170/385) had bronchiectasis ≥1 lobe and 9.4% (36/385) had ≥1 destroyed lobe on HRCT imaging. At 1 year, 30.7% (113/368) reported respiratory symptoms, 19.3% (59/305) and 14.1% (43/305) of patients had experienced declines in FEV1 or FVC of ≥100 mL, 16.3% (62/380) had reported ≥1 acute respiratory event and 12.2% (45/368) had symptoms affecting their ability to work. CONCLUSIONS: PTLD is a common and under-recognised consequence of pTB that is disabling for patients and associated with adverse outcomes beyond pTB treatment completion. Increased efforts to prevent PTLD and guidelines for management of established disease are urgently needed. Low-cost clinical interventions to improve patient outcomes must be evaluated.

Shift-symmetric configurations in two-dimensional cellular automata: Irreversibility, insolvability, and enumeration.

The search for symmetry, as an unusual yet profoundly appealing phenomenon, and the origin of regular, repeating configuration patterns have long been a central focus of complexity science and physics. To better grasp and understand symmetry of configurations in decentralized toroidal architectures, we employ group-theoretic methods, which allow us to identify and enumerate these inputs, and argue about irreversible system behaviors with undesired effects on many computational problems. The concept of so-called "configuration shift-symmetry" is applied to two-dimensional cellular automata as an ideal model of computation. Regardless of the transition function, the results show the universal insolvability of crucial distributed tasks, such as leader election, pattern recognition, hashing, and encryption. By using compact enumeration formulas and bounding the number of shift-symmetric configurations for a given lattice size, we efficiently calculate the probability of a configuration being shift-symmetric for a uniform or density-uniform distribution. Further, we devise an algorithm detecting the presence of shift-symmetry in a configuration. Given the resource constraints, the enumeration and probability formulas can directly help to lower the minimal expected error and provide recommendations for system's size and initialization. Besides cellular automata, the shift-symmetry analysis can be used to study the nonlinear behavior in various synchronous rule-based systems that include inference engines, Boolean networks, neural networks, and systolic arrays.

One Month of Rifapentine plus Isoniazid to Prevent HIV-Related Tuberculosis.

BACKGROUND: Tuberculosis is the leading killer of patients with human immunodeficiency virus (HIV) infection. Preventive therapy is effective, but current regimens are limited by poor implementation and low completion rates. METHODS: We conducted a randomized, open-label, phase 3 noninferiority trial comparing the efficacy and safety of a 1-month regimen of daily rifapentine plus isoniazid (1-month group) with 9 months of isoniazid alone (9-month group) in HIV-infected patients who were living in areas of high tuberculosis prevalence or who had evidence of latent tuberculosis infection. The primary end point was the first diagnosis of tuberculosis or death from tuberculosis or an unknown cause. Noninferiority would be shown if the upper limit of the 95% confidence interval for the between-group difference in the number of events per 100 person-years was less than 1.25. RESULTS: A total of 3000 patients were enrolled and followed for a median of 3.3 years. Of these patients, 54% were women; the median CD4+ count was 470 cells per cubic millimeter, and half the patients were receiving antiretroviral therapy. The primary end point was reported in 32 of 1488 patients (2%) in the 1-month group and in 33 of 1498 (2%) in the 9-month group, for an incidence rate of 0.65 per 100 person-years and 0.67 per 100 person-years, respectively (rate difference in the 1-month group, -0.02 per 100 person-years; upper limit of the 95% confidence interval, 0.30). Serious adverse events occurred in 6% of the patients in the 1-month group and in 7% of those in the 9-month group (P = 0.07). The percentage of treatment completion was significantly higher in the 1-month group than in the 9-month group (97% vs. 90%, P<0.001). CONCLUSIONS: A 1-month regimen of rifapentine plus isoniazid was noninferior to 9 months of isoniazid alone for preventing tuberculosis in HIV-infected patients. The percentage of patients who completed treatment was significantly higher in the 1-month group. (Funded by the National Institute of Allergy and Infectious Diseases; BRIEF TB/A5279 ClinicalTrials.gov number, NCT01404312.).

The Luxembourg Parkinson's Study: A Comprehensive Approach for Stratification and Early Diagnosis.

While genetic advances have successfully defined part of the complexity in Parkinson's disease (PD), the clinical characterization of phenotypes remains challenging. Therapeutic trials and cohort studies typically include patients with earlier disease stages and exclude comorbidities, thus ignoring a substantial part of the real-world PD population. To account for these limitations, we implemented the Luxembourg PD study as a comprehensive clinical, molecular and device-based approach including patients with typical PD and atypical parkinsonism, irrespective of their disease stage, age, comorbidities, or linguistic background. To provide a large, longitudinally followed, and deeply phenotyped set of patients and controls for clinical and fundamental research on PD, we implemented an open-source digital platform that can be harmonized with international PD cohort studies. Our interests also reflect Luxembourg-specific areas of PD research, including vision, gait, and cognition. This effort is flanked by comprehensive biosampling efforts assuring high quality and sustained availability of body liquids and tissue biopsies. We provide evidence for the feasibility of such a cohort program with deep phenotyping and high quality biosampling on parkinsonism in an environment with structural specificities and alert the international research community to our willingness to collaborate with other centers. The combination of advanced clinical phenotyping approaches including device-based assessment will create a comprehensive assessment of the disease and its variants, its interaction with comorbidities and its progression. We envision the Luxembourg Parkinson's study as an important research platform for defining early diagnosis and progression markers that translate into stratified treatment approaches.

Fractalis: a scalable open-source service for platform-independent interactive visual analysis of biomedical data.

Background: Translational research platforms share the aim of promoting a deeper understanding of stored data by providing visualization and analysis tools for data exploration and hypothesis generation. However, such tools are usually platform bound and are not easily reusable by other systems. Furthermore, they rarely address access restriction issues when direct data transfer is not permitted. In this article, we present an analytical service that works in tandem with a visualization library to address these problems. Findings: Using a combination of existing technologies and a platform-specific data abstraction layer, we developed a service that is capable of providing existing web-based data warehouses and repositories with platform-independent visual analytical capabilities. The design of this service also allows for federated data analysis by eliminating the need to move the data directly to the researcher. Instead, all operations are based on statistics and interactive charts without direct access to the dataset. Conclusions: The software presented in this article has a potential to help translational researchers achieve a better understanding of a given dataset and quickly generate new hypotheses. Furthermore, it provides a framework that can be used to share and reuse explorative analysis tools within the community.

Feedforward Chemical Neural Network: An In Silico Chemical System That Learns xor.

Inspired by natural biochemicals that perform complex information processing within living cells, we design and simulate a chemically implemented feedforward neural network, which learns by a novel chemical-reaction-based analogue of backpropagation. Our network is implemented in a simulated chemical system, where individual neurons are separated from each other by semipermeable cell-like membranes. Our compartmentalized, modular design allows a variety of network topologies to be constructed from the same building blocks. This brings us towards general-purpose, adaptive learning in chemico: wet machine learning in an embodied dynamical system.

Risk factors for early mortality on antiretroviral therapy in advanced HIV-infected adults.

BACKGROUND: Many HIV-infected individuals present with advanced HIV disease. These patients are at high risk of death after antiretroviral therapy (ART) initiation, but risk factors for death in these patients are unclear. METHODS: We used data from a multisite randomized trial comparing empiric vs. preventive tuberculosis therapy in HIV-infected adults initiating ART with CD4 T-cell counts less than 50 cells/µl to evaluate risk factors for death within 48 weeks after ART initiation. Cox proportional hazards models were fit to evaluate characteristics present at baseline and at 4 weeks after ART initiation, including the week 4 CD4 T-cell response and new opportunistic infections. RESULTS: Of 850 enrolled, the median pre-ART CD4 T-cell count was 18 cells/µl and 67 (7.9%) died. Baseline risk factors for death included lymphadenopathy, lower CD4 T-cell count, lower serum albumin, high white blood cell count, elevated neutrophil percentage, and lower hemoglobin. Among 746 participants with data at week 4, the median changes in CD4 T-cell count and viral load for those who died (n = 43) vs. survived were 26 vs. 56 cells/µl and -2.7 vs. -2.7 log10 copies/ml, respectively. Each 20 cell/µl lower change in week 4 CD4 T-cell count was associated with a 20% increased risk of post week-4 mortality (adjusted hazard ratio 1.20, 1.01-1.42, P = .038). CONCLUSION: Evidence of active infection and suboptimal immunologic response during the first month of ART are associated with death in the first year after ART initiation in those with advanced HIV disease taking tuberculosis preventive therapy. Strategies to reduce early mortality in this population warrant further investigation.

A nurse-led intervention improves detection and management of AKI in Malawi.

BACKGROUND: Acute kidney injury is common and has significant impact on mortality and morbidity. There is a global drive to improve the lack of knowledge and understanding surrounding the recognition, diagnosis and management of patients with AKI in resource poor healthcare systems. OBJECTIVES: We propose a nurse-led education programme to medical and nursing staff of the Queen Elizabeth Central Hospital (QECH) in Blantyre, Malawi, will improve the overall care and understanding of patients with AKI that will still be effective 3 months later. METHODS: This was a three phase, prospective interventional pilot study which evaluated base line knowledge and clinical practice amongst healthcare workers, provided a comprehensive combination nurse-led class room and ward based teaching programme and evaluated the change in knowledge and clinical management of patients in the high dependency areas of the hospital immediately, and 3 months, after the teaching intervention. RESULTS: The nurse-led intervention significantly improved the healthcare workers attitudes towards detecting or managing patients with suspected AKI (p < 0.0001). There were also significant improvements in the completion of fluid charts and recording of urine output (p < 0.0001), corner stones of AKI management. Knowledge and clinical intervention was still present three months later. There was however little change in the understanding that AKI could be a significant clinical problem in QECH and that it may have a major impact on mortality and working practice and this needs to be addressed in future teaching programmes. CONCLUSIONS: A low cost, nurse-led AKI educational intervention improved the knowledge and management of AKI at QECH, which was still evident 3 months later.

Monocyte unresponsiveness and impaired IL1ß, TNFα and IL7 production are associated with a poor outcome in Malawian adults with pulmonary tuberculosis.

BACKGROUND: Early death during TB treatment is associated with depressed TNFα response to antigenic stimulation and propensity to superadded bacterial infection. Hypothesising the role of monocyte unresponsiveness, we further compared the immunological profile between patients who died or suffered a life-threatening deterioration ('poor outcome') during the intensive phase of TB treatment with patients who had an uneventful clinical course ('good outcome') who had been recruited as part of a larger prospective cohort study of Malawian TB patients. METHODS: Using Luminex, IL1ß, IL2, IL4, IL5, IL6, IL7, IL8, IL10, IL12, IL13, IL17, GCSF, GMCSF, MCP1, MIP1b, IFNγ and TNFα were measured in whole blood assay supernatants (stimulated with Mycobacterium tuberculosis H37Rv and LPS) and serum from 44 Malawian adult TB patients (22 of each outcome) immediately prior to commencing treatment, after 7 days and on day 56 of TB treatment. Monocyte surface expression of CD14, CD16, TLR2, TLR4, CD86 and HLADR, and intracellular TNFα were measured by flow cytometry as was intracellular TNFα response to purified TLR ligands. RESULTS: Lower TB antigen-induced IL1ß (p = 0.006), TNFα (p = 0.02) and IL7 (p = 0.009) were produced in the poor outcome group. TNFα was produced by 'classical' CD14(hi)CD16(lo) monocytes, with no correlation between this response and expression of monocyte surface markers. Response to TB antigens correlated with responses to the purified TLR 2, 3 and 4 ligands. CONCLUSIONS: Dysregulated monocyte cytokine production was identified in TB patients with poor outcome. Lower TNFα responses to H37Rv paralleled lower responses to a panel of TLR ligands, suggesting an underlying perturbation in common TLR signalling pathways. Future work should explore the role of TLR polymorphisms in immune response and clinical outcome in TB patients.

Training an asymmetric signal perceptron through reinforcement in an artificial chemistry.

State-of-the-art biochemical systems for medical applications and chemical computing are application-specific and cannot be reprogrammed or trained once fabricated. The implementation of adaptive biochemical systems that would offer flexibility through programmability and autonomous adaptation faces major challenges because of the large number of required chemical species as well as the timing-sensitive feedback loops required for learning. In this paper, we begin addressing these challenges with a novel chemical perceptron that can solve all 14 linearly separable logic functions. The system performs asymmetric chemical arithmetic, learns through reinforcement and supports both Michaelis-Menten as well as mass-action kinetics. To enable cascading of the chemical perceptrons, we introduce thresholds that amplify the outputs. The simplicity of our model makes an actual wet implementation, in particular by DNA-strand displacement, possible.

Early warning scores generated in developed healthcare settings are not sufficient at predicting early mortality in Blantyre, Malawi: a prospective cohort study.

AIM: Early warning scores (EWS) are widely used in well-resourced healthcare settings to identify patients at risk of mortality. The Modified Early Warning Score (MEWS) is a well-known EWS used comprehensively in the United Kingdom. The HOTEL score (Hypotension, Oxygen saturation, Temperature, ECG abnormality, Loss of independence) was developed and tested in a European cohort; however, its validity is unknown in resource limited settings. This study compared the performance of both scores and suggested modifications to enhance accuracy. METHODS: A prospective cohort study of adults (≥18 yrs) admitted to medical wards at a Malawian hospital. Primary outcome was mortality within three days. Performance of MEWS and HOTEL were assessed using ROC analysis. Logistic regression analysis identified important predictors of mortality and from this a new score was defined. RESULTS: Three-hundred-and-two patients were included. Fifty-one (16.9%) died within three days of admission. With a cut-point ≥2, the HOTEL score had sensitivity 70.6% (95% CI: 56.2 to 82.5) and specificity 59.4% (95% CI: 53.0 to 65.5), and was superior to MEWS (cut-point ≥5); sensitivity: 58.8% (95% CI: 44.2 to 72.4), specificity: 56.2% (95% CI: 49.8 to 62.4). The new score, dubbed TOTAL (Tachypnoea, Oxygen saturation, Temperature, Alert, Loss of independence), showed slight improvement with a cut-point ≥2; sensitivity 76.5% (95% CI: 62.5 to 87.2) and specificity 67.3% (95% CI: 61.1 to 73.1). CONCLUSION: Using an EWS generated in developed healthcare systems in resource limited settings results in loss of sensitivity and specificity. A score based on predictors of mortality specific to the Malawian population showed enhanced accuracy but not enough to warrant clinical use. Despite an assumption of common physiological responses, disease and population differences seem to strongly determine the performance of EWS. Local validation and impact assessment of these scores should precede their adoption in resource limited settings.

Online learning in a chemical perceptron.

Autonomous learning implemented purely by means of a synthetic chemical system has not been previously realized. Learning promotes reusability and minimizes the system design to simple input-output specification. In this article we introduce a chemical perceptron, the first full-featured implementation of a perceptron in an artificial (simulated) chemistry. A perceptron is the simplest system capable of learning, inspired by the functioning of a biological neuron. Our artificial chemistry is deterministic and discrete-time, and follows Michaelis-Menten kinetics. We present two models, the weight-loop perceptron and the weight-race perceptron, which represent two possible strategies for a chemical implementation of linear integration and threshold. Both chemical perceptrons can successfully identify all 14 linearly separable two-input logic functions and maintain high robustness against rate-constant perturbations. We suggest that DNA strand displacement could, in principle, provide an implementation substrate for our model, allowing the chemical perceptron to perform reusable, programmable, and adaptable wet biochemical computing.