AI Insights

| Demo Day Insights | How COVID-19 Pandemic Policies Affected the Vulnerable Populations

June 11, 2020


article featured image

[et_pb_section fb_built=”1″ admin_label=”section” _builder_version=”4.16″ global_colors_info=”{}”][et_pb_row admin_label=”row” _builder_version=”4.19.5″ background_size=”initial” background_position=”top_left” background_repeat=”repeat” hover_enabled=”0″ global_colors_info=”{}” sticky_enabled=”0″][et_pb_column type=”4_4″ _builder_version=”4.16″ custom_padding=”|||” global_colors_info=”{}” custom_padding__hover=”|||”][et_pb_text admin_label=”Text” _builder_version=”4.19.5″ hover_enabled=”0″ z_index_tablet=”500″ text_text_shadow_horizontal_length_tablet=”0px” text_text_shadow_vertical_length_tablet=”0px” text_text_shadow_blur_strength_tablet=”1px” link_text_shadow_horizontal_length_tablet=”0px” link_text_shadow_vertical_length_tablet=”0px” link_text_shadow_blur_strength_tablet=”1px” ul_text_shadow_horizontal_length_tablet=”0px” ul_text_shadow_vertical_length_tablet=”0px” ul_text_shadow_blur_strength_tablet=”1px” ol_text_shadow_horizontal_length_tablet=”0px” ol_text_shadow_vertical_length_tablet=”0px” ol_text_shadow_blur_strength_tablet=”1px” quote_text_shadow_horizontal_length_tablet=”0px” quote_text_shadow_vertical_length_tablet=”0px” quote_text_shadow_blur_strength_tablet=”1px” header_text_shadow_horizontal_length_tablet=”0px” header_text_shadow_vertical_length_tablet=”0px” header_text_shadow_blur_strength_tablet=”1px” header_2_text_shadow_horizontal_length_tablet=”0px” header_2_text_shadow_vertical_length_tablet=”0px” header_2_text_shadow_blur_strength_tablet=”1px” header_3_text_shadow_horizontal_length_tablet=”0px” header_3_text_shadow_vertical_length_tablet=”0px” header_3_text_shadow_blur_strength_tablet=”1px” header_4_text_shadow_horizontal_length_tablet=”0px” header_4_text_shadow_vertical_length_tablet=”0px” header_4_text_shadow_blur_strength_tablet=”1px” header_5_text_shadow_horizontal_length_tablet=”0px” header_5_text_shadow_vertical_length_tablet=”0px” header_5_text_shadow_blur_strength_tablet=”1px” header_6_text_shadow_horizontal_length_tablet=”0px” header_6_text_shadow_vertical_length_tablet=”0px” header_6_text_shadow_blur_strength_tablet=”1px” box_shadow_horizontal_tablet=”0px” box_shadow_vertical_tablet=”0px” box_shadow_blur_tablet=”40px” box_shadow_spread_tablet=”0px” global_colors_info=”{}” sticky_enabled=”0″]

By Devika Bhatia & Laura Clark Murray

A team of 28 AI experts and data scientists collaborated to gauge the impact of pandemic policy implemented post-COVID-19 on vulnerable populations to find correlations and encourage data-driven policymaking to lessen the adversity for the most vulnerable populations around the world.

The entire data analysis including a live demonstration can be found in the demo day recording at the end of the article.

COVID-19 pandemic policy impacting the world’s vulnerable populations

At the onset of the pandemic in 2020, the World Health Organization urged governments to take “urgent and aggressive action” against COVID-19. Many governments reacted with strict measures such as closing borders and quarantining entire cities. Governments all over the world enacted these policies without fully analyzing the factors that impact their effectiveness. Nor did they consider how these policies might deepen the problems for vulnerable populations in different regions.

The project goal: Conduct data-driven impact-analyses on how various pandemic policies affect the well-being of vulnerable populations.

Defining “Vulnerability”

An important step of the project was to define “vulnerability” with respect to the particular context. The project focused on the factors of employment and wage loss, access to health, and domestic violence. To identify the vulnerable population for each of these categories, the team looked to the Inequality-adjusted Human Development Index, considered populations above 65 years of age, and women.

UNDP

Source: UNDP


World maps showing countries by the Inequality-adjusted Human Development Index

World maps showing countries by the Inequality-adjusted Human Development Index


Assessing policies and their effects

The team looked at 17 types of policies from the Oxford COVID-19 Pandemic Government Response Tracker, across the categories of containment, economic response, and health systems. The policies explored included closing of public transportation, stay at home requirements, income support, COVID-19 testing policy, and emergency investment in healthcare.

To analyze the effects of these policies, three key aspects were considered:

  • Time of policy enactment: comparing the time of policy enactment with the effect on a target variable
  • Stringency metric: the degree of intensity of the policies enacted
  • Google Mobility Dataset: quantifies the movement of people in places (e.g. grocery stores vs. parks)

Domestic violence as a ‘Shadow pandemic’

It was ascertained that domestic violence is a growing shadow pandemic as countries displayed a relationship between a decrease in mobility and an increase in the google search rates of relevant topics, coupled with an increase in the number of domestic violence-related articles.

The number of news articles related to both Covid19 and domestic violence started to increase a couple of weeks after the first lockdown measures were implemented in Europe (end of February).

Figure 1: Graph between Ratio of News Articles and Date of recording the values

Figure 1: Graph between Ratio of News Articles and Date of recording the values


The data shows a strong relationship between a decrease in mobility and an increase in Google search rates of domestic abuse topics in many countries. In the countries considered, other than Japan, the peak in search rates has doubled or even tripled, as seen in these graphs of the data from France and India.

Figure 2: Graph between Search trend and mobility change (%) and Date recorded with two different categories namely: Schooling Closing and Workplace Closing for France

Figure 2: Graph between Search trend and mobility change (%) and Date recorded with two different categories namely: Schooling Closing and Workplace Closing for France


Figure 2: Graph between Search trend and mobility change (%) and Date recorded with two different categories namely: Schooling Closing and Workplace Closing for India

Figure 2: Graph between Search trend and mobility change (%) and Date recorded with two different categories namely: Schooling Closing and Workplace Closing for India


The results indicate that the problem of domestic violence could be much bigger than indicated by news stories.

Access to healthcare

The effects of COVID-19 pandemic-response policy measures on access to healthcare, specifically for non-COVID patients was a fascinating angle in this challenge.

The team sought to understand the effects of policy measures on access to healthcare, specifically for non-COVID patients. The vulnerable population was defined based on age, existing chronic medical conditions, and physical access to care facilities. The analysis was focused on England and Wales where there was significant relevant data.

It was found that there was high-mortality among patients with non-COVID chronic diseases during the pandemic as compared to the numbers for the same group in previous years. The data shows a correlation between medical appointment status, such as whether an appointment was kept, changed, or canceled, the stringency of the pandemic policies enacted for the region, and the mobility of the population in that region. In other words, the stringency of pandemic policy and the resulting restrictions on the mobility of a population may cause the medically-vulnerable to miss or avoid regular medical care. And this may be contributing to the increase in non-COVID deaths among this group.

Covid and England-Wales Goverment Policies Impact on Hospitals'Appointments

Covid and England-Wales Goverment Policies Impact on Hospitals’Appointments (Source:Omdena)


The economic impact of pandemic policies

Closures, lockdowns, and decreased mobility have led to wage and employment loss. Though some governments have instituted income support policies, the timing of that aid correlates to employment loss. In countries where income support policies were put in place at roughly the same time as stringent lockdown policies such as workplace closings, the unemployment rate remained relatively flat. This was the case, for example in Sweden and Belgium. In contrast, a delay in the implementation of income support policies correlates to an increased unemployment rate, as was seen in the United States.

Income support policies may affect individuals in the labor force differently. Many countries have undergone employment and wage loss in the informal economy, wherein enterprises, jobs, and workers are not protected by the state.

The team set out to identify the most economically vulnerable populations in this context. The analysis focused on those countries with stringent lockdowns that have implemented income support policies, and in which the population works in sectors highly-impacted by the pandemic policy, such as accommodation and food service, manufacturing, and retail trade.

Some of the results of this analysis are represented here by a mapping of countries according to the stringency of their pandemic policies and the share of their labor force participation in highly impacted sectors. Each country is represented as a circle, the color, and size of which indicates the vulnerability of the workforce in terms of the share of the workforce involved in informal labor.

Vulnerability ranked by Informality Rate

Circle size denotes vulnerability, defined in terms of percentage of worker in high impact sectors and share of workforce involved in informal labor.

Figure 3: The circle size denotes vulnerability, defined in terms of percentage of workers in high impact sectors and share of the workforce involved in informal labor.


This type of topology of the vulnerability of labor forces during the pandemic may be useful in indicating which groups to attend to with income support policies.

Conclusions

While government lockdown policies were designed to slow the spread of COVID-19, they had direct and indirect negative effects on their populations.

  • We found that non-COVID deaths of those with existing health conditions and considerations increased during the pandemic, for the population studied. For this medically-vulnerable population, we found a relationship between the stringency of lockdown pandemic policy and the level of mobility within a locality, with the delivery of non-COVID, and potentially life-saving, healthcare.
  • Domestic violence emerged as a growing “shadow pandemic”. We found a strong relationship between a decrease in mobility of a population and indicators of domestic violence.

To offset the economic impact of anti-contagion policies, many governments instituted income support policies.

  • We determined that the timing of income support policies mattered. For the locations studied, when income support policies were implemented at the same time as lockdown measures, unemployment rates stayed flat. In contrast, in countries where income support policies were delayed, unemployment rate curves remained steep even after policy implementation.
  • The team created economic vulnerability assessments of countries, by considering the stringency of lockdown policies and the share of the labor force involved in highly-impacted sectors and in the informal economy. Income support policies may be more effective when such vulnerability is considered.

Our objective with these results is to support policymakers in finding the most effective ways to minimize the suffering of those most vulnerable.

Find all insights in the demo day recording

[embedyt] https://www.youtube.com/watch?v=H_A2LLA3Bd8[/embedyt]

All Collaborators from this project

We thank our partner organizations, AI for Peace, SH4P, and PWG. as well as all Omdena collaborators (listed below) who made the project a success.

Omdena collaborator

Omdena collaborator

More about Omdena

Omdena is an innovation platform for building AI solutions to real-world problems through the power of bottom-up collaboration.

You might also like:

[/et_pb_text][/et_pb_column][/et_pb_row][/et_pb_section]

Related Articles

media card
An AI Driven Risk Predictor for Mental Health Impacts Due to COVID-19
media card
The Need to Infuse Collaborative Innovation Into the COVID-19 Era
media card
| Demo Day Insights | How COVID-19 Pandemic Policies Affected the Vulnerable Populations