Latest News Archive

Please select Category, Year, and then Month to display items

Developing models to predict patient outcomes in children with severe illness

28 December 2020 | Story André Damons | Photo Supplied

A lecturer from the University of the Free State’s (UFS) department of Paediatrics and Child Health is investigating the use of artificial neural networks to develop models for the prediction of patient outcomes in children with severe illness.

Dr Michael Pienaar, senior lecturer and specialist, is conducting this research as part of his doctoral research and the study deals primarily with the development of models that are designed and calibrated for use in South Africa. These artificial neural networks are computer programs designed to mimic some of the learning characteristics of biological neurons.

The potential applications of models

According to Dr Pienaar these models have traditionally been developed in high-income nations using conventional statistical methods.

“The potential applications of such models in the clinical setting include triage, medical research, guidance of resource allocation and quality control. Having initially begun this research investigating the prediction of mortality outcomes in the paediatric intensive care unit (PICU) I have broadened my scope to patients outside of PICU, seeking to identify children early during their illnesses who are at risk of serious illness requiring PICU,” says Dr Pienaar.

The research up until now has been directed towards the identification of characteristics that are both unique to children with serious illness in South Africa, but also accessible to clinicians in settings where expertise and technical resources are limited.

Research still in the early changes

The research is still in its early stages but next year a series of expert review panels will be held to investigate the selection of variables for the model, after which the collection of clinical data will begin. Once the data has been collected and prepared, a number of candidate models will be developed and evaluated. This should be concluded by the end of 2022.

Says Dr Pienaar: “The need to engage with the rapid proliferation of technology, particularly in the realms of machine learning, mobile technology, automation and the Internet of Things is as great in medical research now as it is in any academic discipline.

“It is critical that research, particularly in South Africa, engage with this in order to take advantage of the opportunities offered and avoid the dangers that go paired with them. Together with the technology as such, it has been essential to pursue this project as an interdisciplinary undertaking involving clinicians, biostatisticians and computer engineers.”

Hope for the research

Dr Pienaar says he was very fortunate and grateful to be the recipient of a generous interdisciplinary grant from the UFS which has allowed him to procure software and equipment that is critical to this project.

“The hope for this research is that the best performing of these models can be integrated with a mobile application that assists practitioners in a wide range of settings in the identification, treatment and early referral of children at high risk of severe illness. I would like to expand this research project to include other countries in Africa and South America and to use it as a bridge to collaboration with other clinical researchers in the Global South,” says Dr Pienaar.

As an early career researcher, Dr Pienaar hopes that this research can serve as a platform to build a body of research that uses the rapid technological advances of these times together with a wide range of collaborations with other disciplines in the pursuit of better child health.

He concludes by saying that he has had excellent support thus far from his supervisors, Prof Stephen Brown (Faculty of Health Sciences, UFS), Dr Nicolaas Luwes (Faculty of Computer Science and Engineering, Central University of Technology) and Dr Elizabeth George (Medical Research Council Clinical Trials Unit, University College London). I have also been supported by the Robert Frater Institute in the Faculty of Health Sciences.

News Archive

Fight against Ebola virus requires more research

2014-10-22

Dr Abdon Atangana
Photo: Ifa Tshishonge

Dr Abdon Atangana, a postdoctoral researcher in the Institute for Groundwater Studies at the University of the Free State (UFS), wrote an article related to the Ebola virus: Modelling the Ebola haemorrhagic fever with the beta-derivative: Deathly infection disease in West African countries.

“The filoviruses belong to a virus family named filoviridae. This virus can cause unembellished haemorrhagic fever in humans and nonhuman monkeys. In literature, only two members of this virus family have been mentioned, namely the Marburg virus and the Ebola virus. However, so far only five species of the Ebola virus have been identified, including: Ivory Coast, Sudan, Zaire, Reston and Bundibugyo.

“Among these families, the Ebola virus is the only member of the Zaire Ebola virus species and also the most dangerous, being responsible for the largest number of outbreaks.

“Ebola is an unusual, but fatal virus that causes bleeding inside and outside the body. As the virus spreads through the body, it damages the immune system and organs. Ultimately, it causes the blood-clotting levels in cells to drop. This leads to severe, uncontrollable bleeding.

Since all physical problems can be modelled via mathematical equation, Dr Atangana aimed in his research (the paper was published in BioMed Research International with impact factor 2.701) to analyse the spread of this deadly disease using mathematical equations. We shall propose a model underpinning the spread of this disease in a given Sub-Saharan African country,” he said.

The mathematical equations are used to predict the future behaviour of the disease, especially the spread of the disease among the targeted population. These mathematical equations are called differential equation and are only using the concept of rate of change over time.

However, there is several definitions for derivative, and the choice of the derivative used for such a model is very important, because the more accurate the model, the better results will be obtained. The classical derivative describes the change of rate, but it is an approximation of the real velocity of the object under study. The beta derivative is the modification of the classical derivative that takes into account the time scale and also has a new parameter that can be considered as the fractional order.

“I have used the beta derivative to model the spread of the fatal disease called Ebola, which has killed many people in the West African countries, including Nigeria, Sierra Leone, Guinea and Liberia, since December 2013,” he said.

The constructed mathematical equations were called Atangana’s Beta Ebola System of Equations (ABESE). “We did the investigation of the stable endemic points and presented the Eigen-Values using the Jacobian method. The homotopy decomposition method was used to solve the resulted system of equations. The convergence of the method was presented and some numerical simulations were done for different values of beta.

“The simulations showed that our model is more realistic for all betas less than 0.5. The model revealed that, if there were no recovery precaution for a given population in a West African country, the entire population of that country would all die in a very short period of time, even if the total number of the infected population is very small. In simple terms, the prediction revealed a fast spread of the virus among the targeted population. These results can be used to educate and inform people about the rapid spread of the deadly disease,” he said.

The spread of Ebola among people only occurs through direct contact with the blood or body fluids of a person after symptoms have developed. Body fluid that may contain the Ebola virus includes saliva, mucus, vomit, faeces, sweat, tears, breast milk, urine and semen. Entry points include the nose, mouth, eyes, open wounds, cuts and abrasions. Note should be taken that contact with objects contaminated by the virus, particularly needles and syringes, may also transmit the infection.

“Based on the predictions in this paper, we are calling on more research regarding this disease; in particular, we are calling on researchers to pay attention to finding an efficient cure or more effective prevention, to reduce the risk of contamination,” Dr Atangana said.

Back

Economic and Management Sciences

Education

Health Sciences

The Humanities

Law

Natural and Agricultural Sciences

Theology and Religion

Open Distance Learning

Business School

We use cookies to make interactions with our websites and services easy and meaningful. To better understand how they are used, read more about the UFS cookie policy. By continuing to use this site you are giving us your consent to do this.

Accept