Chatbot lays down the law for research

WITH Africa known to be burdened more frequently with infectious diseases, a University of KwaZulu-Natal academic has played a leading role in developing an artificial intelligence tool to minimise bureaucracy hindering scientific research.

DATALAW.bot is a chatbot that enables scientists from 12 African countries to navigate with greater ease past the respective data protection laws and ultimately improve health research on the continent.

According to Aliki Edgcumbe, a law lecturer at UKZN, health information is usually classified as special data and inevitably attracts a higher threshold of protection.

Edgcumbe has worked closely with fellow team members at the Data Science for Health Discovery and Innovation in Africa (DS-I Africa) and provided the intelligence for the chatbot that was launched last week.

Edgcumbe’s team is dedicated to providing researchers with legal guidance on how not to be bogged down by the data protection laws in the 12 countries. The AI tool was created to make that process easier.

DS-I Africa is a consortium group that is backed by the National Institute of Health from the US to fund various research projects in Africa where data science is used to conduct health research.

Edgcumbe is also a PhD candidate at UKZN with the focus of her study being data protection laws and how they impact health research in South Africa. The bot was integrated into her academic project.

Edgcumbe and her team usually worked on data sets that already existed and were being reused by scientists for different applications.

“There are usually legal dimensions to it. That’s where we come in to clarify what the law says, how they can use it, new technologies and issues on cross border data sharing.”

Edgcumbe said scientists were required to comply with data sharing legislation relevant to countries they draw information from.

She explained that each country had its own data protection act and data regulations that needed to be adhered to, which made conducting research difficult.

An added challenge was the volume of data researchers needed to handle, especially when dealing with multiple countries.

“The data exchange laws for the various countries are actually very similar, but the devil is in the details.

“Slight differences can have a big impact on outcomes. You want to get it right.

“That’s where the idea for the chatbot is drawn from. We can provide quick responses for scientists to compare the different laws and make a decision.”

Edgcumbe said the bot was trained on specific pieces of legislation from the 12 countries and they ensured the data set was accurate and it churned out responses that were accessible to scientists.

There were instances where researchers waited for as long as three years for ethical clearance before data could be imported into the country for analysis purposes.

“That hampers medical research.

“We want to keep up with the latest trends in science and technology to help people in Africa, especially with learning how diseases are spread, but we can’t do that if data is not shared.

“That was our biggest concern. Our objective is to help scientists to do research instead of sitting and waiting for the all-clear.”

Nations like Nigeria have a list of countries that have equivalent data protection laws and South Africa is on their list.

“But we don’t have our own list, much like many of the countries we deal with.”

She said the bot was able to guide scientists how to use ethical avenues to move data by making certain recommendations.

Edgcumbe said they initially created a guide for scientists to work with data protection laws and made the instructions as readable as possible because legal issues were intimidating for most people.

But with the advent of Chat Gbt, they decided to piggyback off its technology, which allows conversations, when creating their bot.

“It ensured the information was not put across in an intimidating way.”

In creating the intelligence for the bot, Edgcumbe also tackled all the major questions scientists were asking and answered them.

“I worked on the cross-data sharing aspects and streamlined it and I got experts to handle the technical aspects of the bot, before the information was uploaded.

“Then we tested the accuracy and fine-tuned the answering.”

She said as the refining process unfolded she brought in some Masters students and later fourth-year LLB students to assist.

“I trained them on the legislation and what to look for if you were a scientist. They pretended to be scientists and asked the bot multiple questions.

“That helped us to rate its accuracy and usefulness and we flagged responses that were not accurate. We then formulated appropriate responses and fed it back to the bot and in this way we improved its efficiency.”

Edgcumbe feels assured that the bot will help scientists to conduct research with confidence.