Opinionista Co-Pierre Georg 30 July 2020

Covid-19 contact tracing via chatbot? Rather consult a psychic

The health minister’s failure to provide South Africans with a proper digital contact tracing system raises questions about government’s plans for flattening the curve, limiting the spread of the virus and its future management of Covid-19 outbreaks.

“The storm,” to use the words of Minister of Health Dr Zweli Mkhize, “is now arriving.” It hit the Western Cape first before moving on to the Eastern Cape and Gauteng. The big question is: Are we ready for it? Did we buy enough provisions and batten down all hatches? 

I fear the answer is a resounding “no”. There are some provisions we simply couldn’t buy because they are not available — sufficient test kits and ventilators — but some hatches were left wide open by our policymakers. Contact tracing is probably the largest of them. Countries like Germany, with its Corona Warn App which informs users about a potential exposure to Covid-19, are leading the way and have fewer than 1,000 new cases per day. 

Meanwhile, in Mzansi, the Department of Health is trying to sell a chatbot that my students could build as a new tool for contact tracing. COVIDConnect, developed by the Praekelt Foundation and BCX, and recently launched officially by Mkhize, makes it a little easier for users to provide the information of close contacts, but at the cost of the crucial contact tracing interview which can establish the context of an exposure.

We have known since the 1970s that the most potent connections for contagion are “weak links”, for example, when sitting next to a perfect stranger in a taxi filled to capacity. 

COVIDConnect is not a contact tracing tool, but a political stunt by the minister to distract from his previous failure to develop a digital contact tracing system

What is contact tracing and why does it matter? 

Contact tracing is the process of identifying contacts of a person who tests positive for Covid-19 so that the contacts can be notified (either manually or digitally), that they have been exposed to someone who has tested positive and should, consequently self-isolate. The effectiveness of this has been shown in recent work by Oxford researchers, who asked how many contacts of how many symptomatic Covid-19 patients need to be notified in order to curb the spread of the virus. 

The Oxford team finds that timing is of the essence. If there is no delay between the identification of a positive case and the notification of all his/her contacts, then we need to identify about 50% of all positive cases and notify about 60% of her contacts to contain the disease. If we have a two-day delay between identification of a positive case and notification of the contacts, roughly 80% of all contacts need to be notified. 

This is the crux of the matter in South Africa. While the contact tracing process is designed to be largely digital in industrialised countries, South Africa is using an almost entirely manual – and therefore slow – process. Given our country’s low smartphone penetration rate of only about 50%, it is understandable that we do not rely on smartphone technology alone, but alternatives are available. 

Since early March 2020, a team of researchers at the University of Cape Town, together with South African entrepreneurs, built Covi-ID, a free and open-source contact tracing solution specifically designed for users without smartphones. We have sent, literally, thousands of emails to government officials trying to get their buy-in for a privacy-preserving contact tracing solution for South Africa, but there has been no meaningful engagement from either the national or Western Cape Department of Health.

What is worse, the one attempt the South African government had at digital contact tracing – using geolocation data from cellphones via triangulation of the phone’s position – has been an unmitigated disaster and has rightly been scrapped. It was clear from the beginning that the system would not work, because it locates phones only within a 400m radius in an urban setting. In a township, however, there are about 5,000 people within that radius. 

But instead of fixing the issue, the minister is now trying contact tracing via chatbot. Frankly, if he would consult a psychic, he would get more efficient contact tracing. 

No contact tracing – what does it mean? 

If we accept the reality that there is only manual contact tracing in South Africa, what are the consequences? Here are three questions to help us understand the answer. 

First, the most pressing question that everyone – including South African journalists – should ask our policymakers: “Have you accepted that the virus will spread through large parts of our population, or do you intend to contain it?” 

Mkhize earlier spoke about most of the population becoming infected, making it seem inevitable. Implicitly, he has set us on the path of Sweden, but without our government having the guts to step in front of the South African people and explain why they think this is the best way forward. 

Instead, we are being told that we all have to make sacrifices in order to flatten the curve, meaning we sacrifice to give the health system time to build resilience and capacity. Surely, this was necessary, but what Mkhize says is less important than what he does not say. Flattening the curve alone, without other measures, will buy us time, but doesn’t actually solve the problem. 

Other countries show that with the right set of tools, it is possible to contain the virus and we should not accept that the South African government – including the provincial government in the Western Cape – does not pursue the most effective option to actually contain the virus. 

The second question is then: “Are the relaxed lockdown measures effective and do they achieve what they set out to achieve? If not, should we relax restrictions even further, or should we go back to Level 5?” 

The answer to this question strongly depends on the answer to the first question. If we don’t actually plan to “contain” the virus – limit its spread before it reaches a large part of the population – the only reason to go back to Level 5 is to slow down the spread and give our health system time to build capacity. 

Finally, the third important question is: “If we do not intend to contain the virus, how likely are we to see a second or third wave?”

The answer is tricky, because it is intuitive to think that we will be spared future outbreaks because those who have already contracted the virus are unlikely to get it again. 

The problem is that this is far from clear. The virus mutates and, while mutations tend to be more benign (except in movies), we might see future outbreaks of similar viruses – other coronaviruses might become part of our lives. What is more, having a functioning contact tracing system might prove to be a piece of critical infrastructure, similar to a country’s payment system or mobile networks. 

Building an effective national contact tracing solution could be a huge opportunity for the country, not just in the fight against Covid-19, but also to create a thriving new ecosystem for healthcare services in Africa. 

Instead of building infrastructure, we got a glorified chatbot. Poor South Africa. DM/MC



Judge Hlophe’s hand-picked representative from WC Bench to vote on tribunal impeachment verdict

By Marianne Thamm

Beer is drinkable after a nuclear explosion.