The School of Physical and Mathematical Sciences (SPMS) held its fourth School level seminars dubbed 'Afternoons with SPMS'. The seminar, which was held virtually on Thursday, March 17, 2022, was on the topic: “Predicting COVID-19 Cases in Ghana: Exploring Machine Learning Techniques and Time Series Modelling.” It was delivered by Dr. Ezekiel Nii Noye Nortey, Senior Lecturer in the Department of Statistics and Actuarial Science.

In his opening remarks, the Dean of SPMS, Prof. S. M. Yidana reiterated the importance of the seminars, and stated that topics discussed are of national relevance, always eliciting further discussions that provide important data for policymakers. They also foster collaboration among faculty members and industry players.

Dr. Nortey informed the audience that the presentation was a collaborative research work undertaken with Dr. Gabriel Kallah-Dagadu, a Lecturer in the Department of Statistics and Actuarial Science, University of Ghana, and Dr. Theophilus Ansah-Narh, a Research Scientist from the Ghana Space Science and Technology Institute, Ghana Atomic Energy Commission (GAEC).

The presentation started with a general view of the data on the COVID-19 disease, and the number of confirmed cases and fatalities recorded in Africa, as of March 2, 2022. It also looked at Data Description, Time Series Modelling, Artificial Neural Network Approach, Results & Analysis, and Performance Metrics. Dr. Nortey stated that investigating the prevalence of the disease was important to controlling its spread.

Touching on the approach used for the study, Dr. Nortey said the COVID-19 cases in Ghana were analysed using the conventional Time Series Modelling & Artificial Intelligence (AI) approaches. He indicated that globally, 800,000 people are infected by COVID-19 daily. In Africa, the number of COVID-19 cases confirmed is 11,549,076, representing 2.62% of the infections around the world. He said to further understand different attributes of the spread in Africa, the epidemiological metrics such as infection rate, death rate, recovery rate, and ratio were explored. He revealed that with the top 16 countries in Africa, Ghana ranked 15^th in terms of confirmed COVID-19 cases, 12^th regarding recovered cases. In general, Ghana ranked 36^th in death rates in Africa, and 21^st in infection rates in Africa.

To predict the widespread of COVID-19 using Time Series modelling, a model referred to as AutoRegressive Integrated Moving Average (ARIMA) was used. This is a model that integrates autocorrelation measures to model physical structures within the Time Series data to predict future values. Other approaches that could be used are Artificial Neural Networks (ANN), the Long Short Term Memory (LSTM) model networks.

According to Dr. Nortey, being able to estimate new cases and deaths can assist policy makers to plan for cost and facilities needed in managing COVID-19 in the future. He, therefore, solicited for further expansion of this work to investigate the performance of different models. The presentation was followed by a panel discussion and Dr. Nortey was assisted by Dr. Gabriel Kallah-Dagadu. There were questions and contributions on the performance of the various models, the use of machine learning, the data source, and the role of clinical experts in undertaking such a study.  

The seminar attracted about seventy (70) participants including faculty members, staff and students from various units of the University, and from outside the University. The programme was moderated by Dr. Collins Obuah from the Department of Chemistry and was chaired by Prof. Yidana.

Prof. S. M. Yidana (Chairman)

Dr. Collin s Obuah (Moderator)