A Collaborative Approach to the Analysis of the COVID-19 Response in Africa Sharon Okwako1 Irene Wanyana2 Alice Namale2 Betty Kivumbi Nannyonga3

2025-04-28 0 0 1.24MB 6 页 10玖币

侵权投诉

A Collaborative Approach to the Analysis of the

COVID-19 Response in Africa

Sharon Okwako1, Irene Wanyana2, Alice Namale2, Betty Kivumbi Nannyonga3,

Sekou L. Remy1, William Ogallo1, Susan Kizito2, Aisha Walcott-Bryant1, Rhoda Wanyenze2

1IBM Research Africa

Nairobi, Kenya

2Makerere University School of Public Health

3Makerere University School of Physical Sciences

Kampala, Uganda

sharon@ke.ibm.com,iwanyana@musph.ac.ug

anamale@musph.ac.ug,bnk@math.mak.ac.ug

sekou@ke.ibm.com,william.ogallo@ibm.com

kizitosusan@musph.ac.ug,awalcott@ke.ibm.com

rwanyenze@musph.ac.ug

Abstract

The COVID-19 crisis has emphasized the need for scientiﬁc methods such as

machine learning to speed up the discovery of solutions to the pandemic. Har-

nessing machine learning techniques requires quality data, skilled personnel and

advanced compute infrastructure. In Africa, however, machine learning com-

petencies and compute infrastructures are limited. This paper demonstrates a

cross-border collaborative capacity building approach to the application of machine

learning techniques in discovering answers to COVID-19 questions.

1 Introduction

COVID-19 is a global challenge affecting lives worldwide [

]. According to the World Health

Organization, as of

14th

September 2021, there were 225,024,781 conﬁrmed cases of COVID-19

globally, including 4,636,153 deaths [

]. In Africa, there are currently 5,757,213 reported cases

of COVID-19 and 140,002 deaths as of

14th

September 2021 [

]. Given such high mortality and

morbidity rates, countries require appropriate preparedness and response strategies to control disease

spread and limit deaths [

]. There is also a need for timely and informed decision-making in

implementing interventions and allocating resources, especially in low and middle-income countries

with heavily burdened and fragile health systems [5].

With the rapidly evolving COVID-19 pandemic and the urgent need to combat the disease, data ag-

gregation, and analysis are critical for timely, effective, and efﬁcient decision-making. Unfortunately,

the exponential increase in COVID-19 related data has rendered traditional data analysis techniques

inefﬁcient [

], thus requiring automated tools to extract hidden insights from massive data within

the shortest time possible. Machine learning and artiﬁcial intelligence have been utilized to offer

innovative solutions to public health concerns [

]. They provide tools to support disease intervention

planning, especially during a pandemic like COVID-19 [

]. Several researchers have developed

machine learning, and artiﬁcial intelligence techniques to aid the healthcare sector and policymakers

forecast COVID-19 disease trends and diagnose and treat COVID-19 [9, 10].

In Africa, however, human resource and infrastructural capacities for machine learning and artiﬁcial

intelligence are still very minimal. Furthermore, the increasing amount of health-related data is not

equivalent to the available resources and techniques for its utilization [6].

35th Conference on Neural Information Processing Systems (NeurIPS 2021), Sydney, Australia.

arXiv:2210.01882v1 [cs.LG] 4 Oct 2022

This project aimed to assess the COVID-19 pandemic responses in Africa and to build skills for

machine learning within University Institutions in Africa. The institutions included Makerere

University School of Public Health (Uganda) which was the lead university, the University of Ibadan

in Nigeria, Cheikh Anta Diop University (Senegal), and the Kinshasa School of Public Health

(Democratic Republic of Congo) in collaboration with IBM-Research Africa. These universities have

a close working relationships with the Ministries of Health stakeholders in their respective countries

and would provide insights generated from the project to policy makers for decision-making.

2 Methodology

2.1 Research and Technical Collaboration

The overarching research goal of this collaboration is to assess the impact of and response to the

COVID-19 pandemic in East, Central, and West Africa. To this end, the collaboration is anchored on

a multimodal program of research with several objectives such as (1) to document the government

policies and response strategies to COVID-19, (2) to evaluate the effect of the response strategies

(e.g. non-pharmaceutical intervention (NPI) measures such as lockdowns, mask-wearing, and

social distancing) on the control of COVID-19, (3) to evaluate the effect of COVID-19 and related

interventions on essential non-COVID care, (4) to evaluate the preparedness of health systems in

handing COVID-19 and related interventions, and (5) to develop context-relevant strategies that

inform policy and decision-making. To date, the collaboration has conducted the studies in the

Democratic Republic of Congo, Nigeria, Senegal, and Uganda, with plans underway to including

more partner countries.

To generate rich insights for decision-making, we combine traditional qualitative and quantitative

research methods with state-of-the-art machine learning techniques for analyzing real-world evidence

data. For example, to assess the impact of the COVID-19 pandemic on essential healthcare in Uganda,

we used a mixed methods approach involving qualitative in-depth interviews (IDIs) and quantitative

interrupted time series analyses (ITS). The qualitative IDIs aimed to describe the key informant

perspectives on barriers to continuity of essential health service delivery. This investigation revealed

several context-speciﬁc themes such as abandonment of critical policies, movement restrictions

affecting health workers, suspension of outreach and support activities, and fear of fear of infection

among patients. The quantitative ITS aimed to evaluate the impact of the introduction of COVID-19

NPIs on new clinic visits, diabetes clinic visits, and newborn hospital deliveries across the Central,

Eastern, Northern, and Western regions of Uganda. As exempliﬁed in Figure 1, we found out that new

clinic visits declined in the Northern, Central, and Western regions, and appear to have increased in

the Eastern region. Furthermore, to amplify the value of COVID-19 population-based mathematical

models (compartmental models), we applied model calibration based on artiﬁcial intelligence that

incorporates information about the stringency of government responses to COVID-19 [11].

The core research team of our collaboration was comprised of an interdisciplinary cross-functional

team of domain experts, research scientists, and software engineers, and business analysts working

closely together and guided by agile values and principles. Throughout the collaboration, the team

met weekly to discuss progress towards set goals and milestones, as well to discuss insights generated

from the team’s endeavors. Key decisions e.g., research questions to pursue, methodologies to

use, and training and capacity building needs where made through collaborative consultations. To

overcome the overheads of collaboration across different geographies, the team primarily met via

online web conferencing applications.

2.2 Training, Dissemination and Reuse

One key objective of this collaboration was establishing and strengthening a network/community of

practice of universities focused on assessing the impact of public health emergencies of international

concern in general and COVID-19 in particular for policy makers, based on existing networks housed

by Makerere University School of Public Health (MakSPH). To this end, IBM Research-Africa

planned a number of virtual capacity building sessions to build participant’s interest and capacity in

data analysis and interpretation of results using available tools and resources. Some of the speciﬁc

objectives of the sessions included teaching participants how to: 1) Use of tools such as python

for machine learning 2) Prepare data and perform data analysis in python using examples from

the MakSPH COVID-19 project 3) Interpret data analysis results and visualizations. The training

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

ACollaborativeApproachtotheAnalysisoftheCOVID-19ResponseinAfricaSharonOkwako1,IreneWanyana2,AliceNamale2,BettyKivumbiNannyonga3,SekouL.Remy1,WilliamOgallo1,SusanKizito2,AishaWalcott-Bryant1,RhodaWanyenze21IBMResearchAfricaNairobi,Kenya2MakerereUniversitySchoolofPublicHealth3MakerereUniversitySchoolo...

展开>> 收起<<

A Collaborative Approach to the Analysis of the COVID-19 Response in Africa Sharon Okwako1 Irene Wanyana2 Alice Namale2 Betty Kivumbi Nannyonga3.pdf

共6页,预览2页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

A Collaborative Approach to the Analysis of the COVID-19 Response in Africa Sharon Okwako1 Irene Wanyana2 Alice Namale2 Betty Kivumbi Nannyonga3

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: