How AI and Apps are Revolutionising Medical Diagnoses

If an HR department was able to table up the number of hours employees spend checking on whether a mole is skin cancer or a rash is psoriasis, heads in management would likely explode.

Finding medical info online has been – on the whole – revolutionarily helpful. For some, it’s comforting to read that your condition, “generally resolves by itself within a week.” But for those predisposed to hypochondria, the internet is exacerbating fears, turning mild ailments into life-threatening possibilities. The rabbit holes are deep. There are so many awful things to learn about!

Type “headache” in a search box and you’ll get millions of results, which will include every possibility from mild dehydration to a brain aneurism. What’s needed is some winnowing. “Headache” is too generic a word for even the best search engines. More data is required. But what if you aren’t all that clear about how to describe your own symptoms? What if you need some promoting? There’s an app for that, and it runs on artificial intelligence (AI).

The K-Health app runs on a solid premise: use big data and AI to compare cases, ask the right kind of questions, make a correct guess, and suggest treatment. The K-Health website explains it thus: “K asks smart questions that account for your age, gender, and medical history as it investigates your health. After a short text-based conversation, K will share how doctors have diagnosed and treated people with similar symptoms and backgrounds.” And there is a second angle that’s attracting many, “If K recommends it, you can instantly chat with a doctor with the click of a button.”

So far, K-health reviews have been positive; with the company claiming over 10,000 five-star reviews on app stores. Those offering negative comments are more likely to complain about the human doctors used by the app than the AI wizard – called “K”. The AI – according to a study cited by the app’s parent company – gets the same diagnosis as a physician an impressive 85 percent of the time. And the beauty of AI is that it never stops trying to improve.

Much ink has been spilled bemoaning the state of healthcare in the United States, with good reason. Elections can be swayed over who has the better healthcare plan and U.S. voters often rank healthcare at the top of their concern list. Lacking “socialized” medicine, many Americans are reluctant to consult a doctor due to fees. It’s hard to even give a range of pricing for what seeing a doctor in the U.S. might cost. Those with insurance might pay US$50, or more, or less. But the un- or underinsured can expect bills of several hundred dollars for basic care. For too many, it’s literally a question of food and transport this month, or a check-up for that cough that won’t go away. CNBC reports that over last year, “22% of Americans say they have steered clear of some sort of medical care — including doctor visits, medications, vaccinations, annual exams, screenings, vision checks, and routine blood work — because of the expense.”

The app’s value lies in it being affordable, on-demand, personalized, and fast. The AI – in seconds – compares millions of medical records (without knowing the identity of those patents) to show how doctors have diagnosed similar cases based on similar symptoms. The AI probes with questions gleaned from a user’s replies, and is able to pick up cues for more in-depth inquires by assessing responses. Of course, at its core, AI is math. In a process that can take less than two minutes, K-Health offers patients percentages. “Based on (number of) similar cases to yours, the likelihood of X is X.”

Check-ups for symptoms are free to all downloaders. Those who decide to subscribe, pay for a function that allows users to chat with live doctors 12 hours-a-day, seven-days-a-week. This formula is something AI will be doing a lot of in the near future: acting as a baseline for a specialist to review. Users are offered unlimited doctor chats at one price level or a flat fee for a single consultation.

The biggest downside is vision. The AI might correctly detect that you have strep throat, but being unable to look down your actual throat, could miss something important. Users are sometimes asked to take photos of their conditions for a doctor to review, but there are those who aren’t willing to try getting a phone camera picture of the inside of their throat.

Imaging will improve, but more importantly, so will the AI. Still in its infancy, K-Health is still figuring out exactly where it’s strongest. Mental health, for example, seems a perfect candidate for AI. A turbulent political environment, the Covid-19 pandemic, lockdowns, isolation and a general climate of fear are the makings of mental health meltdowns. Untold numbers of people across the world are going to need help coming to terms with this new reality, but the reality is – there just aren’t enough trained professionals to help them. Financial concerns and a shortage of providers will likely push patients toward the only feasible solution: Dr. AI.

Patients thrive when a strong personal relationship or inter-personal chemistry is formed with a mental health professional. Too many, however, fruitlessly search for that person that works for them. Chatting with an AI shrink takes personality out of the encounter, removes worries about costs as you watch the clock and the anonymity might prompt more truthful answers. In fact, some argue an algorithm be may be superior to a live professional.

Founded in 2016, K-health now has a reported US$97 million in funding; a healthy amount for R&D. Meanwhile, the company is building alliances in the industry, linking up with major U.S. medical insurers – and their data. The app is also reportedly working on more language support, with Spanish a top priority. K-health may not be perfect, but it’s such a massive leap forward that it’s not hard to imagine a future where a patient would decide to take the advice of a mega-data-connected AI over the judgment of a flesh-and-blood doctor.