Issue link: https://maltatoday.uberflip.com/i/1242414
30.04.2020 8 OPINION T he Chinese government built the most effective epidemic control system ever created in the histo- ry of the world using advanced technol- ogies. In recent weeks, the pandemic in China was under control, with only 50 cases reported daily. For a country, having 1.4 billion in- habitants, this is a resounding success. However, the system was widely crit- icised by people in the western world since it requires access to private infor- mation about the whereabouts of Chi- nese people; their chats, social contacts, their purchases and additional informa- tion. is data is then stored in central databases and used to fight the spread of the virus with advanced Artificial In- telligence (AI) algorithms. In the following article, we will pro- pose a similar system which is as effec- tive but which addresses the privacy concerns of the individual. e person- al data of the individual will never leave the person's device, but it will still help the centralised system identify potential hotspots. e only requirement of such a sys- tem is that people install an app on their mobile device. With difference to the Chinese system, it doesn't require an army of video surveillance cameras dis- tributed around the country or biome- tric scanners installed in the doorways of residential complexes. People do not need to check-in or out of their living quarters either. e system takes the form of a mo- bile application, similar to the WeChat which exists in China, and it also inte- grates different services. So localisation, social media, chats and e-wallets will all be available. However, this is where the similarities end. Rather than having a centralised server, the user's phone stores all of the personal information. e user will have full access to the data, and he can choose to retain or delete it. Furthermore, the system will use two AI components, one located re- motely and the other on the device per se, as will be explained below. In the case of the e-Wallet, the first advantage is that payments occur using contactless virtual cards and without exchanging real money. us, this lack of physical exchange automatically re- duces the dissemination of the virus. e e-wallet also includes information about shops visited together with the purchases. e system stores the loca- tion of the shop and logs the time in the user's device. is piece of information is then processed at a later stage by the localisation module. From the purchases, the local AI can gather information regarding the well- being of the individual. It can easily in- fer that if certain items (such as medi- cines) are purchased, then the person or someone close to him is most probably ill. e AI then starts a chat with the user, enquires whether he or someone close to him has some symptoms and proposes a way forward. ese can include beginning self-quar- antine, making an appointment to be swabbed or taking any measure which deemed appropriate. e important thing is that the choice remains in the hands of the users, and no information gets shared with any department. Such an AI system is possible because today, we have AI engines which efficiently work on a mobile device without having to communicate with a server. One such tool is TensorFlow Lite specifically de- signed for on-device machine learning. Furthermore, the applications of mo- bile-medicine are on the rise. ere are various symptom-tracker apps which help users make appropriate decisions on whether they need institutionalised care or not. It also provides symptom relief for minor illnesses which they can handle on their own. If the user requires specific assistance, it can connect him directly with the appropriate provider. Médicins sans Frontières even went a step further and created the first Mobile Triage App. Carnegie Mellon Universi- ty went a step further and just launched the first COVID Voice Detector which claims to test if the user has the virus, only by talking through his mobile phone. Of course, this system is still ex- perimental, and they give no guarantee on its accuracy. e localisation module provides pre- cise information regarding the wherea- bouts of the person (within an error of a few meters). Of course, this informa- tion is only stored on the device and not shared with anyone. We can use this information with- out compromising the user's identi- ty through a novel AI approach called federated learning. e system works as follows; the device downloads the AI model from the server and improves it locally by learning from the data on the phone. A summary of the new model is created and sent to the cloud. No personal information ever leaves the device, and it is impossible to ex- tract any information from the update sent because it is just a summary. With the AI model on the device, the system knows which areas of the city are dan- gerous. If the user comes in close con- tact to a location which had someone infected with the virus, the AI alerts him immediately and gets him to avoid that area. rough the social media module, if a user gets infected, he can easily send a warning message to his social circle of friends and alert them so that they can take the necessary precautions. e system can also automatically cal- culate a Health code which can be ei- ther; red, amber or green. is coding is there to help users make choices which protect the safety of others. A green colour means that the person is free to roam unrestricted. People who just re- turned from abroad or might have been in an infected area would have a yellow code, and the system will advise them to limit their mobility for a few days. ose who are probably infected fall under the red category and they will have to stay in quarantine. e scope of such an app is not there to replace the manual system but to in- form people and guide them into taking appropriate actions. It also relieves the stress from the healthcare system by leveraging over the power of AI to reach more people. In the end, no system is perfect and achieving a balance is not easy. However, by combining the benefits of the Chinese system with the priva- cy standards expected in most western countries, we will have a new powerful tool which can save countless lives au- tomatically. Alexiei Dingli Prof Alexiei Dingli is a Professor of AI at the University of Malta and has been conducting research and working in the field of AI for more than two decades, assisting different companies to implement AI solutions. He forms part of the Malta.AI task-force, set up by the Maltese government, aimed at making Malta one of the top AI countries in the world Tweaking China's coronavirus tech strategy to beat the virus in the West