Page 181 - AI for Good-Innovate for Impact Final Report 2024

AI for Good-Innovate for Impact

as 90% of people with CKD are unaware that they have impaired kidney function, because it
is usually asymptomatic until it reaches an advanced stage.

This project aims to develop an AI system that would be able to aggregate all the collected
data at patient level, process it through data cleaning and feature engineering, develop and 41-CU&CH
train a model on such data to predict the likelihood of developing CKD or, if already present,
the likelihood to get worse next year. Through the use of Explainable AI techniques, e.g. SHAP
or Causal SHAP, we will then be able to understand each individual's main risk factors, to then
intervene on those people to help them take the right actions to reduce their risk, leading to
health and financial improvements.

UN Goals:
• SDG 3: Good Health and Well-being
• SDG 10: Reduced Inequality
• SDG 17: Partnerships to achieve the Goal

Justification of UN Goals selection: SDG 3: Good Health and Well-being: The primary goal
of this project is to predict chronic kidney disease (CKD), thereby contributing directly to
SDG 3, which focuses on ensuring healthy lives and promoting well-being for all at all ages.
By leveraging advanced predictive analytics, the project aims to enable early detection and
intervention for CKD, ultimately improving patient outcomes and reducing the burden of
disease on individuals and healthcare systems. SDG 10: Reduced Inequality: Chronic Kidney
Disease often disproportionately affects certain populations due to inequalities based on
ethnicity, gender, and socioeconomic status. Addressing SDG 10, which seeks to reduce
inequality within and among countries, this project emphasizes the need for an ethical and
bias-free approach in its predictive models. By ensuring that the predictive tools are inclusive
and equitable, the project aims to mitigate disparities in CKD diagnosis and treatment, fostering
a more just healthcare system. SDG 17: Partnerships for the Goals: The project's collaborative
framework aligns with SDG 17, which advocates for strengthening the means of implementation
and revitalizing the global partnership for sustainable development. Carna's industry expertise,
combined with connections to academic experts, underscores the value of multi-sectoral
collaboration. This synergy not only enhances the project's impact but also showcases the
potential of partnerships in driving innovative solutions for global health challenges.

Partner name: Carna Health Partner

41�2�2 Future work

Proof of concept development, Model development, Collection of more data

41�3 Use-case requirements

• REQ-01: It is required that there is a comprehensive collection of patient screening data,
medical history, medication records and demographic information.
• REQ-02: It is critical that there are secure databases for storing patient data and data
integration tools to handle diverse data sources.
• REQ-03: It is crucial that there is development of a highly accurate machine learning
model to make appropriate predictive analysis.
• REQ-04: It is critical that there are methods to visualize and communicate individual risk
factors for predictive analysis interpretability.

165