AI for increased equality between the sick and the healthy

With the rise of new scientific discoveries such as immunotherapies or CRISPR gene modification, humanity is increasing its influence on probabilities and timing of the onset of diseases. However, the speed of time remains the only thing we can’t influence in life. We can only manage the time we have. In healthcare, time is managed poorly today. China is one of the countries whose startup ecosystem began to bridge the gap with the help of AI-supported healthcare.

Waiting time is increasing the gap between healthy and chronically ill.

In many cases, patients need to wait for months for an appointment with specialists and even then, it can quickly happen that hours pass in waiting rooms before they see a doctor. This is consequently further increasing the gap between the healthy and those with chronic conditions because chronic patients’ quality of life is not only hindered by the disease itself, but also additional disease-related costs and time lost in the healthcare system.

This is especially problematic in public systems. China is a good example: there, primary care is underused and public hospitals are severely understaffed and overburdened. Patients don’t trust primary care physicians on local levels so they travel to cities to see specialists in hospitals. They can wait up to several days, for only a few minutes long consultations. With the rise of big data and progress hungry entrepreneurs, driven by the promises of AI, the solutions for these issues are starting to take shape. 

AI-supported healthcare to bridge the gap.

Ping An Good Doctor is a Chinese startup, providing an AI-supported one-stop healthcare ecosystem platform in China. According to the company, the platform is used by 265 million registered users already. The solution enables patients to get a piece of medical advice, first by a triage with an AI-supported bot that collects their medical history and adds s preliminary diagnostic suggestion. The screening is followed by a consultation with a real doctor to ensure accuracy. The company has an internal team of 1000 doctors and contracts with over 5000 doctors to provide 24/7 support for patients. Ping An Good Doctor also started setting up 1-minute clinics across China – kiosks for medical advice with safely stored and refrigerated 100 categories of common drugs, potentially enabling patients to avoid the need to visit a pharmacy. 

Ping An Good Doctor was designed by over 200 AI specialists on a dataset of 400 million consultations.

This is a clear example of why China has a strong advantage in AI development – China’s population exceeds that of Europe and the USA combined, consequently producing large numbers of AI scientists and offering large data sets of patients. 

Yitu is another successful Chinese company whose scientists were the first to co-author a scientific paper in Nature about using natural language processing (NLP) to achieve high accuracy rates when reading electronic health records and generating patient diagnoses. Yitu has 400 full time and part-time doctors that work on labeling data. The dataset for the system described in Nature included 1.3 million outpatient visits and 100 million data points.

Why some countries fall behind.

Because of the quantity of data in China, and the way they approached its use, China has ideal means for rapid progress in AI. Quality results with AI require large quantities of quality data. Smaller populations could potentially fall behind in development due to the lack of financial interest to develop well designed natural language processing models for them. In this sense, the Chinese market is large enough to drive the financial interest in new products and services. 

In the last few years, FDA approved more than a dozen algorithm for improved diagnostics. Insurance companies are looking at ways to support patients with chatbots, they are developing driver performance monitoring algorithms and insurance market analytics algorithms. Healthcare providers are exploring potential increases in efficiency, for example, Amazon partnered with the Boston-based Beth Israel Deaconess Medical Center to use machine learning to reduce delays and rescheduling of patient procedures.  They are exploring the use of AI for better workforce schedules planning by assessing the overall level of risk in intensive care units and predicting when the hospital will experience an unexpectedly high volume of incoming patients.  

To support the right decisions, we need high-quality data.

One of the biggest dangers in AI supporting decisions in healthcare is, that if algorithms are based on poor data, they could produce harmful suggestions, consequently harming large numbers of patients. Scientists also worry about the potentially harmful consequences of the misalignment of values between people and AI in times of Superintelligence or General Artificial Intelligence (GAI) – AI that would be capable of human-like understanding. Predictions regarding GAI span from 2050 to 2200

In the end, technology is just technology. It’s our decision and responsibility to use it wisely and decide to what extent we are going to use it for decision support and where it will take over decision making. If the efficiency in healthcare does improve with the help of AI, in fields of scheduling, diagnostics, and disease management, this could be a beginning of increased equality between those that get sick and those that never set their foot in the healthcare system. This could have exponentially positive consequences for society, as patients would spend less time in the healthcare system more time pursuing their dreams and ambitions. 

Guest post by Tjasa Zajc, Business Development and Communications Manager at Better by Marand and Data Natives 2019 speaker.


If you would like to find out more about the gamechanging role of AI in healthcare and meet Tjasa Zajc, get your DN19 ticket here.


Up Next....


What is blockchain technology used for?

Since its First launch, blockchain technology has been associated with the world of cryptocurrencies and because of this, it has raised certain doubts. The lack of transparency of some virtual