Sempa JB Welte A.

PloS one

Introduction

Viral Load (VL) monitoring is a crucial component of patient care during antiretroviral therapy (ART) but is not routinely available in many resource-constrained settings, where millions of patients will require care for decades to come. We hypothesise a serologic 'recent infection' test (Sedia LAg assay) which has a high dynamic range for detecting antigen-driven antibody response can provide informative proxies for VL trajectories.

Methods

A retrospective study where we analysed data linked via specimens in a well-described repository for recent infection test benchmarking (CEPHIA collaboration). Patient panels were comprised of 1) observations straddling ART start; 2) observations from a period of stable viral suppression; 3) observations straddling rebound after a period of viral suppression. We analysed an individual's Sedia LAg ELISA normalised optical density (ODn) trends within these categories. Using groups 2) and 3) we evaluated the specificity and sensitivity of a proposed proxy for "the latest observation is at a time of VL rebound"; proxy was defined as follows: we estimated patient-specific mean-previous-ODn for all observations with at least two preceding virally suppressed observations. We considered various thresholds to define both "VL suppression" and "ODn uptick".

Results

In regression analysis by category: 1) ODn gradients are statistically significantly negative just after ART-start (p = 0.010); 2) During periods of stable viral suppression, ODn tended to decline, but not statistically significantly, for a range of clinically meaningful "VL suppression" thresholds; 3) comparing ODn values just before, versus at, "VL rebound", ODn changes were statistically significantly increasing at rebound (p = 0.001). In the analysis comparing groups 2) and 3), at a Z score threshold of 0.8, the proposed proxy for a first viral rebound had an observed specificity and sensitivity both close to 90%.

Conclusion

The high dynamic range of serological tests previously investigated for defining 'recent infection' has potential, as demonstrated using the Sedia LAg ELISA, to provide meaningful information about the success of ART, during treatment initiation, at times of stable suppression, and to flag possible viral rebound. It should be investigated how this can be combined with patient management workflows and (clinical and) other data, to provide efficiencies in long-term monitoring viral control in resource-limited settings.

2024-06-06

Pienaar MA Sempa JB Luwes N George EC Brown SC.

Frontiers in pediatrics

Objectives

Delays in identification, resuscitation and referral have been identified as a preventable cause of avoidable severity of illness and mortality in South African children. To address this problem, a machine learning model to predict a compound outcome of death prior to discharge from hospital and/or admission to the PICU was developed. A key aspect of developing machine learning models is the integration of human knowledge in their development. The objective of this study is to describe how this domain knowledge was elicited, including the use of a documented literature search and Delphi procedure.

Design

A prospective mixed methodology development study was conducted that included qualitative aspects in the elicitation of domain knowledge, together with descriptive and analytical quantitative and machine learning methodologies.

Setting

A single centre tertiary hospital providing acute paediatric services.

Participants

Three paediatric intensivists, six specialist paediatricians and three specialist anaesthesiologists.

Interventions

None.

Measurements and main results

The literature search identified 154 full-text articles reporting risk factors for mortality in hospitalised children. These factors were most commonly features of specific organ dysfunction. 89 of these publications studied children in lower- and middle-income countries. The Delphi procedure included 12 expert participants and was conducted over 3 rounds. Respondents identified a need to achieve a compromise between model performance, comprehensiveness and veracity and practicality of use. Participants achieved consensus on a range of clinical features associated with severe illness in children. No special investigations were considered for inclusion in the model except point-of-care capillary blood glucose testing. The results were integrated by the researcher and a final list of features was compiled.

Conclusion

The elicitation of domain knowledge is important in effective machine learning applications. The documentation of this process enhances rigour in such models and should be reported in publications. A documented literature search, Delphi procedure and the integration of the domain knowledge of the researchers contributed to problem specification and selection of features prior to feature engineering, pre-processing and model development.

2023-02-21

Evbuomwan O Van Rensburg BJ Engelbrecht G Driver CHS Labuschagne M Sempa J Horn-Lodewyk J.

Nuclear medicine and molecular imaging

Purpose

Our objectives were to investigate the utility of

^99m

Tc-ethylenedicysteine-deoxyglucose (ECDG) in identifying active disease in the joints of patients with rheumatoid arthritis (RA), as well as to evaluate the biodistribution of this radiopharmaceutical.

Methods

A prospective study was conducted at the Department of Nuclear Medicine of the University of the Free State/Universitas Academic Hospital in Bloemfontein, South Africa. Twenty-two participants from the rheumatology department diagnosed with RA according to the ACR/EULAR classification criteria were enrolled. Participants were injected with 20-25 mCi of

^99m

Tc-ECDG. Flow, blood pool, whole body, delayed static, and SPECT/CT images were acquired. Known sites of disease were qualitatively assessed for intensity of uptake, and disease severity was graded (Grade 0-3).

Results

Twenty-two participants were studied. The median (interquartile range) age was 59 (49-68) years, and the majority (

n

 = 21; 95.5%) were females. There was abnormal increased uptake of

^99m

Tc-ECDG noted in majority of the sites of known disease, including unknown sites. SPECT/CT imaging localized radiotracer uptake specifically to the synovial space. Similar biodistribution of radiotracer was noted in all patients, irrespective of disease severity or fasting status.

Conclusion

^99m

Tc-ECDG can efficiently assess disease activity in the joints of patients with RA. It accumulates in sites of both clinical and subclinical disease and might be a very useful tool for the rheumatologist in the management of patients with RA.

2023-09-25

Sempa JB Patil R Mathewson JD Kabelka H Yaghmaei N Coleman H Sohoni P Straetemans M Gopalakrishna G Wienia M Kombe F Alba S.

BMJ global health

Introduction

In the past decade, global health research has seen a growing emphasis on research integrity and fairness. The concept of research integrity emerged in response to the reproducibility crisis in science during the late 2000s. Research fairness initiatives aim to enhance ownership and inclusivity in research involving partners with varying powers, decision-making roles and resource capacities, ultimately prioritising local health research needs. Despite extensive academic discussions, empirical data on these aspects, especially in the context of global health, remain limited.

Methods

To address this gap, we conducted a mixed-methods study focusing on research integrity and fairness. The study included an online frequency survey and in-depth key informant interviews with researchers from international research networks. The dual objectives were to quantify the frequency of practices related to research integrity and fairness and explore the determinants influencing these practices in global health.

Results

Out of 145 participants in the quantitative survey (8.4% response rate), findings indicate that global health researchers generally adhere to principles of research integrity and fairness, with variations in reported behaviours. The study identified structural, institutional and individual factors influencing these patterns, including donor landscape rigidity, institutional investments in relationship building, guidelines, mentoring and power differentials among researchers.

Conclusion

This research highlights that, despite some variations, there is a substantial alignment between research integrity and fairness, with both sharing similar determinants and the overarching goal of enhancing research quality and societal benefits. The study emphasises the potential to explicitly recognise and leverage these synergies, aligning both agendas to further advance global health research.

2024-03-22