Abstract:
Human immunodeficiency virus (HIV) is the virus responsible for the condition called
Acquired Immunodeficiency Syndrome (AIDS). Since the first report of HIV in the early
1980s, its adverse effects on humanity cannot be overemphasised. HIV infection has led to
the loss of many lives and adversely affected businesses across the globe. As a result, it
has had very significant negative impacts on the world‟s economy in many adverse ways.
Sub-Saharan Africa (SSA) is known to be most affected by this virus; accounting for
approximately 70 per cent of the world‟s HIV infected population. Annual death records
and new infections caused by the virus are highest in SSA. Attempts by Researchers,
Governments and intergovernmental bodies have been largely successful in combating its
spread but not enough to eliminate it since SSA remains a region of a generalised
epidemic. Many researchers have identified fundamental Socio-economic drivers of the
virus in the Sub-Saharan African context in various population subgroups. These drivers
are usually identified as a list of prevailing factors characterising the spread of the virus in
specific communities without further details regarding their measures of impact. Minimal
efforts have been made in an attempt to assess, in quantitative measures, the contribution
of individual drivers to the spread of the virus. Furthermore, not much has been done in an
attempt to investigate possible interdependencies or interactivities among such Socio economic drivers. Such details, if known, would give policymakers a deep insight into the
behaviour of such drivers. With detailed insights, the objective to halt further new
infections of the virus could be achieved much faster. Ghana is the contextual setting for
this research. Ghana is situated in Sub-Saharan Africa where several institutions have been
set up to formulate policies and to roll out campaigns to combat the spread of the virus.
This research develops a data-driven computational model for assessing the degrees of
impact and interdependences of Socio-economic HIV drivers using Feature Maximization,
which is an emerging data mining technique. Feature Maximization first splits the dataset
into several clusters before assessing the effect (degree of impact) of each driver contained
in the dataset. This approach gives equal opportunity for the impact of each driver to be
assessed adequately. This approach makes it possible to report each Socio-economic HIV
driver together with its degree of impact for the given context of the study, which was not
the case in earlier works. It is clear from the outcome that, the degrees of impact and
interactivities of a given HIV driver depends on the setting (rural or urban), education
level, marital status and occupation. The results obtained shows that, low (formal) education and disrupted marital statuses such as divorce, separation and widowhood are
strong drivers of the epidemic. Rural males of age groups 40-44 and females of age groups
30-34 and 40-49 are most prone to the epidemic. In the urban settings, the male age group
most infected are 30-34 and 50-54 while the most infected age groups for females are 25-
29, 35-39 and 40-44. Elementary occupations such as Crafts and Related Trades Workers
were found to be strongly associated with the epidemic in urban areas while technicians
and associate professionals are most at risk of getting infected by the rural setting. The
developed model runs in linear time and is therefore suitable for large datasets. It would be
very useful for stakeholders and policymakers in their quest to curb the HIV epidemic. It
is recommended that future research works devise ways to establish the direction of
causality among HIV drivers.