1 of 7

Sauki.ai: Modelling AI Use Cases in Public Health

Prepared By:

Corona Management Systems

4 February 2024

2 of 7

Sauki is an AI tool that enhances science communication by making public health modelling research easier to understand and use for African audiences

Overview of Sauki

  • Summarise modelling research articles
  • Trained with briefs from health and disease modelling research papers from across 9 African countries
  • It was designed with templates that ensure fidelity to the information provided in the article
  • Extracts and preprocesses content from complex research papers
  • It also tailors the summaries to meet the needs of different audiences by including sections relevant to each group and using simplified language that aligns with their specific interests and expertise

AI: Artificial Intelligence;

Source: Corona Management System Team analysis

2

10 December 2024

3 of 7

Sauki is easy to use and the process for generating briefs is simple – just create an account, upload an article and select your preferred brief type

Visit the site

Go to sauki.ai

Create an Account

Use your email address to create an account and create a simple password

Upload your article

Log in and select an article in PDF format and upload by clicking at the left top of the page

Select Audience type

Select your audience type (Donors, Academics, General Public, or Decision Makers)

Create a brief

Select Generate Brief and wait for about 90 seconds for the brief to be generated

Submit Feedback

Select Submit Feedback and input your feedback to help us improve the tool or send it via email

AI: Artificial Intelligence; PDF: Portable Document Format

Source: Corona Management System Team analysis

3

10 December 2024

4 of 7

Sauki.ai helps to bridge key gaps to help everyone understand public health research, irrespective of your background

Gap Between Research and Funding Decisions for Donors

Sauki briefs for donors will help to ensure better understanding of health insights with which to drive decisions on resource allocation

Gap Between Research and Policy

Policymakers are now able to generate briefs tailored to them, helping to better drive evidence-based policymaking and ensure governance decisions are well-informed

Gap in Non-Specialists' Understanding of Research

With Sauki, you do not have to be a public health specialist or a modeller to understand and stay up to date with complex health research

Gap in Access to Summarised Evidence

Public health professionals often need quick, reliable summaries of complex information to inform decision-making and interventions

Barriers to Cross-Sector Collaboration

Public health advancements can be communicated more effectively to everyone from modellers to researchers outside of public health, making inter-disciplinary research collaborations even easier 

Source: Corona Management System Team analysis

4

10 December 2024

5 of 7

Sauki was built using standardised templates derived through a human-centered design process ensuring that the specific needs of all user types are considered

Human-Centred Design

Standardised Templates

Sauki was co-designed along with potential users of the tool who fit into the different audiences being considered to improve user experience

This approach helped in the creation and validation of templates for the briefs to suit each audience type

The standard templates include general sections, along with additional sections tailored to each audience, ensuring the use of appropriate technical language to meet their specific needs.

This approach, in addition to prompt engineering, also ensures briefs are produced with only information from the uploaded paper leaving Sauki with no chance to analyse therefore no information is lost or added

Responsible AI Practices

There were ethical considerations while the tool was built such as HCD, transparency, data protection, inclusiveness, alignment with ethical standards, as well as risk management

AI: Artificial Intelligence; HCD: Human-Centred Design

Source: Corona Management System Team analysis

5

10 December 2024

6 of 7

Modellers formed a core part of the development of templates and validation of the briefs generated by the tool

6

05 February 2025

Identification of modelling articles

Modellers provided guidance on how to identify a modelling research paper

Validation of brief template

Modellers co-created the template for the briefs to ensure it effectively captured every aspect of the research paper

Validation of human-generated briefs

They read the human-generated briefs and made corrections that were necessary before the briefs were used to train the LLM

Validation of Sauki-Generated briefs

After deployment of the tool, modellers used it to generate briefs and gave their feedback on the output

LLM: Large Language Model

Source: Corona Management System Team analysis

7 of 7

Thank you!