MediAccess: AI Healthcare Data Analysis and Prediction

AI Healthcare Data Analysis and Prediction

Introduction

In a world where healthcare systems must manage their pressures to provide better care outcomes with limited resources, data-based insights are more important now than they have ever been before. Being able to analyze and predict trends in the healthcare sector can allow policymakers, healthcare providers, and researchers to be able to make decisions that improve public health so that this allows for a better performant system to be created for the population. 

One such work that is being done right now is MediAccess AI. A data analysis and machine learning project that uses healthcare data from developing countries to identify trends and make predictions. Utilizing a dataset from the World Bank Group Health Data, this analysis dives into important healthcare indicators such as population growth, life expectancy, and fertility rates. In this blog, we will be looking into the analysis, the revelations that have been discovered, and how this work can inform healthcare policies and interventions.

Healthcare systems in Sub-Saharan Africa and Nigeria are faced with unique challenges that are closely linked to scarce resource availability and increasing population growth. 

The Dataset

The dataset used in this project spans from 1960 to 2023 and includes a wide range of healthcare indicators. It includes a wide range of indicators such as:

• Population Growth: How populations in developing countries have changed over time.
• Life Expectancy: Trends in life expectancy, a key indicator of healthcare quality.
• Fertility Rates: Changes in fertility rates that have significant implications for population dynamics and healthcare planning.

The dataset had 67,564 rows and 69 columns. The dataset was cleaned and preprocessed before analysis was done.

I focused on Sub-Saharan Africa and Nigeria, and the key indicators include:

• Population Growth: How populations in Sub-Saharan Africa and Nigeria have changed over time.
• Life Expectancy: Trends in life expectancy, a key indicator of healthcare quality.
• Fertility Rate: Changes in fertility rates that have significant implications for population dynamics and healthcare planning.

Data Cleaning and Preprocessing

This is the foundation of reliable analysis. Before diving into the analysis, the data had to be cleaned and prepared for modelling. The analysis and predictions are on GitHub

• Handling Missing Values: Missing data points were replaced with 0.
• Dropping Unnecessary Columns: Columns like "Unnamed: 68" were removed to streamline the dataset and focus on the most relevant indicators.

This preprocessing step was crucial to ensuring the accuracy and reliability of the analysis.

Exploratory Data Analysis (EDA)

The main reason for the exploratory data analysis is to uncover trends and patterns. The data was visualized to uncover trends and patterns. Here are some key insights:

Subsaharan Africa and Nigeria

• Population Trends in Sub-Saharan Africa and Nigeria

Population growth was analyzed in Sub-Saharan Africa and Nigeria. The data showed a rapid increase in population from 1960 to 2023. This trend is consistent with global patterns of population growth, but it’s important to note that Sub-Saharan Africa and Nigeria have some of the highest population growth rates in the world.

Population growth in Sub-Saharan Africa and Nigeria from 1960 to 2023

• Life Expectancy in Sub-Saharan Africa and Nigeria

Life expectancy is a key indicator of healthcare quality. In Sub-Saharan Africa and Nigeria, life expectancy has shown a consistent upward trend over the years, but it still lags behind global averages. This is likely due to challenges such as limited access to healthcare services, high rates of infectious diseases, and poor living conditions.

Life expectancy in Sub-Saharan Africa and Nigeria from 1960 to 2023

•  Fertility Rates in Sub-Saharan Africa and Nigeria

Fertility rates in Sub-Saharan Africa and Nigeria have declined over the past few decades, but they remain among the highest in the world. This trend has significant implications for population growth and healthcare planning, as high fertility rates can strain healthcare systems and resources.

Fertility rates in Sub-Saharan Africa and Nigeria from 1960 to 2023

Trends for Nigeria

Other countries like Aruba.

• Population Trends in Aruba

One of the first things analyzed was the population growth in Aruba. The data shows a steady increase in population from 1960 to 2023, with a few fluctuations. This trend is consistent with global patterns of population growth, but it’s important to note that Aruba’s population growth has slowed in recent years.

Population growth in Aruba from 1960 to 2023

• . Life Expectancy in Aruba

Life expectancy is a key indicator of healthcare quality. In Aruba, life expectancy has shown a consistent upward trend over the years. This is likely due to improvements in healthcare infrastructure, access to medical services, and better living conditions.

Life expectancy in Aruba from 1960 to 2023.

• Fertility Rates in Aruba

Fertility rates have declined significantly in Aruba over the past few decades. This trend is consistent with global patterns, where fertility rates tend to decrease as countries develop and access to healthcare and education improves.

Fertility rates in Aruba from 1960 to 2023

Trends for Aruba

PowerBI Analysis

The data was analyzed with PowerBI to uncover insights on life expectancy,  Adolescent fertility rate, birth rate and death rate, as well as mortality rate and survival. Here is the link to access the dashboard

Machine Learning: Predicting Healthcare Trends

With the dataset of data having been cleaned and analyzed, the next critical step was to construct machine learning models to predict future healthcare trends. You can find the saved algorithms for the different countries, the visualizations, and Jupyter Notebook on the GitHub page. For this, two algorithms were used:

• Random Forest Regressor: An ensemble learning method is particularly well suited for regression tasks.
• XGBoost: A gradient-boosting algorithm that is known for its performance and speed.

Model Evaluation

The models had scores determined by evaluation using mean squared error (MSE) and R-squared (R2) scores. Both models performed satisfactorily; however, XGBoost outperformed RandomForestRegressor in terms of accuracy by a slight margin.

Random Forest's Feature Importance

XGBoost's Feature Importance

Random Forest: Actual vs Predicted

XGBoost: Actual vs Predicted

Key Insights from the Analysis

Subsahara Africa

• Population growth is rapid. The population in Sub-Saharan Africa, including Nigeria, is rapidly growing; thus, this could likely have an effect on healthcare systems and resources. There is now an increase in the need to access healthcare, requiring access to vaccines and medical equipment to cater to an increase in the population; therefore, policymakers need to plan for an increased demand for healthcare services in both urban and rural areas.
• Life expectancy is increasing, but still below global averages. In Sub-Saharan Africa and Nigeria, life expectancy has increased; however, it still remains significantly behind markers of global averages. This highlights the need for continued investment in healthcare infrastructure and services.
• Fertility rates are high. In Sub-Saharan Africa and Nigeria, fertility rates continue to remain in the highest position in the world. This has extensive consequences for population growth and healthcare planning as high rates of fertility cause a high demand for mothers' and children's services.

Why This Is Important: Implications for Healthcare Policy.

The insights produced from the analysis have major implications for healthcare policy as well as planning in Sub-Saharan Africa and Nigeria.

• Resource Allocation: There is a need to understand population trends, especially when it comes to health care. Understanding the population trends is key to the effective allocation of resources by policymakers. An example of this can be seen in the situation where rapid population growth will require more investment to be made in the healthcare infrastructure and services.
• Healthcare Infrastructure: The increase in the life expectancy rate suggests that the healthcare infrastructure is improving, but there is still sufficient room for further growth, even in rural areas.
• Family Planning: High fertility rates highlight the importance of family planning services, which are capable of ensuring that population growth is sustainable in the long run.

The MediAccess-AI project shows strengths in both data science and machine learning for the purpose of healthcare. By looking at historical information, it is possible to uncover patterns that are important to policy decisions as well as improvements to healthcare outcome prediction. Being able to predict future healthcare needs based on history is a major change and especially advantageous when resources are limited in areas such as Sub-Saharan Africa and Nigeria.