Indepth Network

Modern AI systems rely on complex infrastructure stacks involving data pipelines, machine learning operations (MLOps), and cloud services. However, centralized control of these components introduces risks related to data manipulation, lack of traceability, and governance failures. This article presents a decentralized AI infrastructure model where blockchain is integrated into MLOps workflows to enhance data governance, model traceability, and lifecycle transparency. Smart contracts are used to enforce governance policies, control access to datasets, and register model artifacts throughout their lifecycle. The proposed architecture demonstrates how blockchain can complement existing AI infrastructure by providing decentralized coordination, auditability, and policy enforcement without compromising performance or scalability.

In the contemporary hyper-competitive business landscape, the ability to transform raw data streams into actionable insights with minimal delay is a primary driver of strategic advantage. This research paper evaluates the systemic impact of Azure Event Hub–based streaming architectures on enterprise decision latency the temporal gap between a business event and the subsequent corrective or opportunistic action. By examining the integration of high-throughput ingestion via Azure Event Hubs, real-time transformation through Azure Stream Analytics, and continuous state synchronization via Change Data Capture (CDC), this study articulates a framework for reducing data and analysis latencies. Furthermore, the inclusion of Artificial Intelligence for IT Operations (AIOps) is analyzed as a mechanism for autonomous system maintenance and incident remediation. Drawing upon empirical benchmarks and industry success stories, including Netflix’s anomaly detection frameworks and Penguin Random House’s inventory optimization, the paper demonstrates that transitioning from traditional batch-oriented Lambda architectures to unified, streaming-first Kappa architectures can reduce decision latency by 17–25%. The findings suggest that such architectures not only improve operational key performance indicators (KPIs) by 9–14% but also fundamentally transform the role of the corporate strategist from an intuitive visionary to an empirical architect.

ObjectivesApproximately 80% of people with epilepsy live in low- and middle-income countries (LMICs), where limited resources and stigma hinder accurate diagnosis and treatment. Clinical machine learning models have demonstrated substantial promise in supporting the diagnostic process in LMICs by aiding in preliminary screening and detection of possible epilepsy cases without relying on specialised or trained personnel. How well these models generalise to naïve regions is, however, underexplored. Here, we use a novel approach to assess the suitability and applicability of such clinical tools to aid screening and diagnosis of active convulsive epilepsy in settings beyond their original training contexts.MethodsWe sourced data from the Study of Epidemiology of Epilepsy in Demographic Sites dataset, which includes demographic information and clinical variables related to diagnosing epilepsy across five sub-Saharan African sites. For each site, we developed a region-specific (single-site) predictive model for epilepsy and assessed its performance at other sites. We then iteratively added sites to a multi-site model and evaluated model performance on the omitted regions. Model performances and parameters were then compared across every permutation of sites. We used a leave-one-site-out cross-validation analysis to assess the impact of incorporating individual site data in the model.ResultsSingle-site clinical models performed well within their own regions, but generally worse when evaluated in other regions (p<0.05). Model weights and optimal thresholds varied markedly across sites. When the models were trained using data from an increasing number of sites, mean internal performance decreased while external performance improved.ConclusionsClinical models for epilepsy diagnosis in LMICs demonstrate characteristic traits of ML models, such as limited generalisability and a trade-off between internal and external performance. The relationship between predictors and model outcomes also varies across sites, suggesting the need to update specific model aspects with local data before broader implementation. Variations are likely to be particular to the cultural context of diagnosis. We recommend developing models adapted to the cultures and contexts of their intended deployment and caution against deploying region- and culture-naïve models without thorough prior evaluation.

South Africa has the largest share of people living with HIV in the world and this population is ageing. The social context in which people seek HIV care is often ignored. Apart from clinical interventions, socio-behavioural factors impact successful HIV care outcomes for older adults living with HIV. We use cross-sectional data linked with demographic household surveillance data, consisting of HIV positive adults aged above 40, to identify socio-behavioural predictors of a detectable viral load. Older adults were more likely to have a detectable viral load if they did not disclose their HIV positive status to close family members (aOR 2.56, 95% CI 1.89-3.46), resided in the poorest households (aOR 1.98, 95% CI 1.23-3.18), or were not taking medications other than ART (aOR 1.83, 95% CI 1.02-1.99) likely to have a detectable. Clinical interventions in HIV care must be supported by understanding the socio-behavioural barriers that occur outside the health facility. The importance of community health care workers in bridging this gap may offer more optimum outcomes for older adults ageing with HIV.

Abstract ObjectivesApproximately 80% of people with epilepsy live in low- and middle-income countries (LMICs), where limited resources and stigma hinder accurate diagnosis and treatment. Clinical machine learning models have demonstrated substantial promise in supporting the diagnostic process in LMICs without relying on specialised or trained personnel. How well these models generalise to naïve regions is, however, underexplored. Here, we use a novel approach to assess the suitability and applicability of such clinical tools for diagnosing active convulsive epilepsy in settings beyond their original training contexts.MethodsWe sourced data from the Study of Epidemiology of Epilepsy in Demographic Sites dataset, which includes demographic information and clinical variables related to diagnosing epilepsy across five sub-Saharan African sites. For each site, we developed a region-specific (single-site) predictive model for epilepsy and evaluated its performance on other sites. We then iteratively added sites to a multi-site model and evaluated its performance on the omitted regions. Model performances and parameters were then compared across every permutation of sites. We used a leave-one-site-out cross-validation analysis to assess the impact of incorporating individual site data in the model.ResultsSingle-site clinical models performed well within their own regions, but worse in general when evaluated on other regions (p&lt;0.05). Model weights and optimal thresholds varied markedly across sites. When the models were trained using data from an increasing number of sites, mean internal performance decreased while external performance improved.ConclusionsClinical models for epilepsy diagnosis in LMICs demonstrate characteristic traits of ML models, such as limited generalisability and a trade-off between internal and external performance. The relationship between predictors and model outcomes also varies across sites, suggesting the need to update specific aspects of the model with local data before broader implementation. Variations are likely to be specific to the cultural context of diagnosis. We recommend developing models adapted to the cultures and contexts of their intended deployment and caution against deploying region- and culture-naïve models without thorough prior evaluation.Key pointsMachine learning-driven clinical tools are becoming more prevalent in low-resource settings; however, their general performance across regions is not fully established. Given their potential impact, it is crucial models are robust, safe and appropriately deployedModels perform poorly when making predictions for regions that were not included in their training data, as opposed to sites that wereModels trained on different regions can have different optimal parameters and thresholds for performance in practiceThere is a trade-off between internal and external performance, where a model with better external performance usually has worse internal performance but is generally more robust overallSEEDS collaboratorsAgincourt HDSS, South Africa: Ryan Wagner, Rhian Twine, Myles Connor, F. Xavier Gómez-Olivé, Mark Collinson (and INDEPTH Network, Accra, Ghana), Kathleen Kahn (and INDEPTH Network, Accra, Ghana), Stephen Tollman (and INDEPTH Network, Accra, Ghana)Ifakara HDSS, Tanzania: Honratio Masanja (and INDEPTH Network, Accra, Ghana), Alexander MathewIganga/Mayuge HDSS, Uganda: Angelina Kakooza, George Pariyo, Stefan Peterson (and Uppsala University, Dept of Women’s and Children’s Health, IMCH; Karolinska Institutet, Div. of Global Health, IHCAR; Makerere University School of Public Health), Donald NdyomughenyiKilifi HDSS, Kenya: Anthony K Ngugi, Rachael Odhiambo, Eddie Chengo, Martin Chabi, Evasius Bauni, Gathoni Kamuyu, Victor Mung’ala Odera, James O Mageto, Isaac Egesa, Clarah Khalayi, Charles R NewtonKintampo HDSS, Ghana: Ken Ae-Ngibise, Bright Akpalu, Albert Akpalu, Francic Agbokey, Patrick Adjei, Seth Owusu-Agyei, Victor Duko (and INDEPTH Network, Accra, Ghana)London School of Hygiene and Tropical Medicine: Christian Bottomley, Immo KleinschmidtInstitute of Psychiatry, King’s College London: Victor CK DokuUCL Queen Square Institute of Neurology, London: Josemir W SanderSwiss Tropical Institute: Peter Odermatt

We provide here the first bottom-up review of the lived experience of depression, co-written by experts by experience and academics. First-person accounts withinand outside the medical field were screened and discussed in collaborative workshops involving numerous individuals with lived experience of depression, family members and carers, representing a global network of organizations. The material was enriched by phenomenologically informedperspectives and shared with all collaborators in a cloud-based system. The subjective world of depression was characterized by an altered experience of emotions and body (feeling overwhelmed by negative emotions, unable to experience positive emotions, stuck in a heavy aching body drained of energy, detached from the mind, the body and the world); an altered experience of the self (losing sense of purpose and existential hope, mismatch between the past and the depressed self, feeling painfully incarcerated, losing control over one's thoughts, losing the capacity to act on the world; feeling numb, empty, non-existent, dead, and dreaming of death as a possible escape route); and an altered experience of time (experiencing an alteration of vital biorhythms, an overwhelming past, a stagnation of the present, and the impossibility of the future). The experience of depression in the social and cultural context was characterized by altered interpersonal experiences (struggling with communication, feeling loneliness and estrangement, perceiving stigma and stereotypes), and varied acrossdifferentcultures, ethnic or racial minorities, and genders. The subjective perception of recovery varied (feeling contrasting attitudes towards recovery, recognizing recovery as a journey, recognizing one's vulnerability and the need for professional help), as did the experience of receiving pharmacotherapy, psychotherapy, and social as well as physical health interventions. These findings can inform clinical practice, research and education. This journey in the lived experience of depression can also help us to understand the nature of our own emotions and feelings, what is to believe in something, what is to hope, and what is to be a living human being.

Background: Urinary schistosomiasis caused by infection withSchistosoma haematobium (S. haematobium) remains endemic in Africa and is associated with haematuria and albuminuria/proteinuria. Kidney Disease Improving Global Outcomes clinical guidelines recommend evaluating proteinuria/albuminuria and glomerular filtration rate for chronic kidney disease (CKD) diagnosis. The guidelines are informed by population data outside of Africa but have been adopted in many African countries with little validation. Our study aimed to characterise the burden of urinary schistosomiasis in rural South Africa (SA) and evaluate its relationship with markers of kidney dysfunction with implications for CKD screening.Methods: In this population-based cohort study, we recruited 2021 adults aged 20 – 79 years in the Mpumalanga Province, SA. Sociodemographic data were recorded, urinalysis performed, and serum creatinine and urine albumin and creatinine measured. Kidney dysfunction was defined as an estimated glomerular filtration rate (eGFR) <60ml/min/1.73m2 and/or urine albumin-creatinine ratio >3.0mg/mmol. S. haematobium infection was determined by urine microscopy. Multivariable analyses were performed to determine relationships betweenS. haematobium and markers of kidney dysfunction.Results: Data were available for 1226 of 2021 participants. 717 (58.5%) were female and the median age was 35 years (IQR 27 – 47). Prevalence of kidney dysfunction andS. haematobium was 20.2% and 5.1% respectively.S. haematobium was strongly associated with kidney dysfunction (OR 8.66; 95% CI 4.10 – 18.3) and related to albuminuria alone (OR 8.69; 95% CI 4.11 – 18.8), with no evidence of an association with eGFR <90ml/min/1.73m 2 (OR 0.43; 95% CI 0.05 – 3.59).Discussion: The strong association between urinary schistosomiasis and albuminuria requires careful consideration when screening for CKD. Screening for, and treatment of, schistosomiasis should be a routine part of initial work-up for CKD inS. haematobium endemic areas. Urinary schistosomiasis, a neglected tropical disease, remains a public health concern in the Mpumulanga province of SA.

Abstract Over the past 18 years, the industry has focused on drilling and completing horizontal producing wells, in order to develop better deliverability and UER across the reservoir. These types of wells are drilled with various inclinations, trajectories which porpoise across the reservoir interval. Horizontal lateral sections consist of numerous selectively stimulated stages containing greater than 20 stages spread out across 2,000 ft to 25,000 ft of horizontal interval. (Heddleston, 2017).Many indirect measurements such as chemical tracing or fiber optic sensing have been trialed, but the only direct measurement of real production contributing across a producing lateral is acquired using production logging measurements with production data analysis. This true measurement of production contribution is the best method which allow oil companies to understand and acknowledge which portions of the hydraulically stimulated reservoir is contributing the production and the value of their properties.