Billions of people use Google to find health-related topics and to get diagnosed. But the answers delivered by advanced search engines don’t necessarily correspond to evidence-based medicine. Will symptoms checkers and personal health guides substitute “Dr Google”? Do pre-diagnosis apps have the potential to reorganize the healthcare market?
Evidence-based medicine is an AI-based medicine
The need to self-diagnose is not an invention of the 4th industrial revolution. Before the era of internet and smartphones, people used books to acquire medical knowledge, match symptoms with diseases and find related medicines and treatment tips. Health encyclopedias in the home library stock were the exclusive source of reliable information.
The Internet and free access to the knowledge changed everything. In 1996 internet entrepreneur Jeff Arnold founded WebMD, one of the top healthcare websites with 75 million unique visitors monthly and 52 million unique mobile users. Since then, many medical websites have tried to meet the needs of patients: Those who are concerned about their health and look for answers online, who have no access to medical services or fight with stigmatized health disorders and are afraid to visit a doctor. Over the last 10 years the way people can acquire knowledge has changed drastically. For example, WikiDoc, launched in 2006, was “the first and largest wiki dedicated to medicine on the internet with an initial focus on cardiovascular disease”.
As a consequence of technological development, health online encyclopedias evolved towards advanced expert websites covering most popular health topics and offering self-diagnosis tools. The National Health Service created so-called “self-help guides” on the most relevant health topics like flu, headaches, lower back pain etc. After answering a few questions, patients get advice or are redirected to the doctor if required. After entering a postcode, local medical services are suggested. Choose a symptom, select related factors and view possible causes – the Mayo Clinic (USA) uses a similar online tool to the NHS. After a short questionnaire it displays a list of diseases and conditions that match at least one of the selected factors.
In 2008 Apple presented its first mobile phone. Since then, information is available whenever and wherever it’s needed, including for healthcare-related issues. In 2019 the number of mobile phone users is forecast to reach 4.68 billion. In 2016 almost 2/3 of the world population owned a mobile phone. Developers very quickly recognized a new niche: AI-based doctors in a pocket in a mobile form of symptoms checkers and chatbots. The promise is to revolutionize healthcare, make it affordable, address the problem of unreliable diagnoses made by “Dr Google”, improve the access to the medical services in rural areas, and empower patients. Entrepreneurs from all over the world have the ambitious mission of breaking the monopoly of Google as the first source of health-related topics.
The problem seems to be serious: Medical fake news on the internet is a rising concern for health systems. Researchers from BUPA found that 47 percent of patients who use popular internet search engines to diagnose themselves, on the first page of results see at least one result for cancer. Dr Haider Warraich, a fellow in heart failure and transplantation at Duke University Medical Center, in an article published in The New York Times “Dr. Google Is a Liar” claims that fake medical news threatens our democracy and our lives. What’s more, when patients make wrong health-related decisions, the consequences might be expensive. Consequently healthcare is becoming unsustainable and limited resources are being wasted.
Another aspect of AI-based health assessment is to support medial professionals. “One in every seven diagnoses is incorrect. 1.5 million people around the world die each year due to misdiagnosis. It poses a threatening burden for healthcare systems, payers, and the industry,” says Dr Jama Nateqi, Co-Founder of Symptoma, a website where patients and doctors enter symptoms to receive a list of matching causes sorted by probability. No one doctor can have the knowledge about all of 20,000 diseases, AI can get and process this knowledge and support doctors in the decision-making process.
“The goal of Your.MD is to democratise healthcare and place a doctor in the pocket of every citizen in the world” – declares Your.MD – an AI health information service. Due to its “Pre-Primary Care Report”, benchmark tests have shown medical accuracy at 85% for the 20 most common conditions. Medical accuracy for a further 500 conditions now sits at more than 60%.
Multifunctionality to cover all patients’ needs
Mobile apps for diagnostics are becoming multi-tasking platforms offering expert medical advice, allowing users to search for a doctor or to find appropriate over-the-counter medicines. They follow the fast expanding digital health trend that, in the last few years, led to the liberalization of medical services provision in many countries around the world. Face-to-face doctors’ visits are being supplemented by telemedicine services. Doctor apps – like Babylon and Kry – are offering 24/7 access to consultations. Both of the latter companies already have more than 5 million users around the world. Asia has adapted to this change in healthcare much faster than Europe or the USA. The world’s biggest online doctors’ platform is China’s “Ping An Good Doctor” with 228 million registered users (June 2018), 531,000 consultations in the first half of 2018 and over 5,600 employed or outsourced doctors (1,037 in-house medical professionals and 4,650 external doctors).
The digital health market is set to exceed $379 billion by 2024.
Among doctors’ apps, symptom checkers promise something else: Fast, anonymous, free-of-charge health assessments. Moreover, engaging Artificial Intelligence in a simple diagnosis – where doctors’ expertise is not required – also tackles the problem of rising medical staff shortages. Estimates are alarming: 400 million people worldwide lack access to essential health services, 6% of people in developing countries are impoverished as a result of having to pay for the health care they need.
Strategy behind technology
One of the reasons why the solutions like Ada or Your.MD have become so popular is the enormous market potential. According to a new report by Global Market Insights, the digital health market is set to exceed $379 billion by 2024. Just in the last few years, Artificial Intelligence, algorithms and Big Data have made it possible to create solutions able to draw conclusions based on symptoms. That’s the reason why most companies in this field are startups that entered the market after 2014.
“Our technology is based on probabilistic modeling and a set of statistical algorithms that improve over time. In general, we learn from two primary sources – medical experts and machine learning. First of all, we hire qualified and experienced doctors who use available evidence-based literature to provide the initial information about conditions, symptoms, risk factors and their probabilistic relations. This is handled by the authoring tool that we have developed called MetaBase. Until now we have spent about 10,000 hours of physician review,” explains Piotr Orzechowski, CEO of Infermedica that offers an online symptom checker called Symptomate. “Secondly, we use available structured datasets to analyze and extract additional information that can augment information provided by the doctors. Importantly, the suggestions provided by machine learning algorithms are also curated and reviewed by our doctors. We have currently analyzed clinical notes for more than 1 million hospital patients, 1 million outpatient visit summaries and about 3 million interviews with users of our symptom checker,” explains Orzechowski.
Symptomate is a suite of web, mobile and voice apps that help patients assess their symptoms when they’re feeling unwell. The goal is to analyze symptoms and recommend the most suitable next steps. The solution covers 13 language versions including English, German, Spanish, French, Arabic, Chinese and many others. The symptom checker is available in three major voice platforms: Amazon Alexa, Microsoft Cortana and Google Assistant. “The audience includes mainly young adults and more than 55% of the users are aged 18 – 30. Some of the most commonly reported symptoms include headaches, back pains, mild abdominal pains and overall fatigue,” says Piotr Orzechowski.
With a total funding of $67m, ADA Health is another big player on the market. The ADA app has 6 million users and 10 million assessments completed, it’s available in 5 languages and ranked as the #1 medical app in over 130 countries. The Berlin-based company employs 130 workers and over 40 doctors and medical editors.
Recently, Ada has announced a partnership with one of the German health insurers Techniker Krankenkasse (TK). Users of the Ada app , who are insured in the TK, can enter their complaints/health symptoms, receive a personal analysis and will be informed on request about suitable digital care offers of TK.
To check if this patient service approach would add a value to a traditional model, Ada ran a trial with NHS GP clinics. The results are promising. “92% of GPs said it was useful to receive an Ada assessment before a consultation. 65% said it saved them time; on average, two minutes of a seven minute consultation. In another trial run with a large NHS GP clinic, 14% of patients that completed an Ada assessment in the waiting room said that if they had used Ada at home they would not have felt the need to come to see the doctor that day,” says Claire Novorol, founder of Ada Health in an interview published in Sifted (February 2019). Every company offering health tech apps emphasizes that the aim is not to replace medical professionals but to support them.
Solutions like Ada or Symptomate have to fit into the complex healthcare ecosystem and constantly expand their functionality in accordance with patients’ expectations and market trends.
“We are partnering worldwide with health providers, organisations and other businesses, where there is a lot of potential for us to add value. Some of the partners we are already working with are the Techniker Krankenkasse, Sutter Health and the NHS”, summarizes Vincent Jörres, PR- and Communications-Manager in Ada Health.
Public health providers also decide to partner with pre-diagnosis apps developers, for example the National Health Service. To relive the overloaded 111 non-emergency consultation services, NHS launched the “Ask NHS – Virtual Assistant” app (powered by Sensely). A virtual assistant named Olivia ask questions about the symptoms and suggests what the problem is. If needed, Olivia arranges a call back from a 111 nurse to discuss the symptoms further. Patients can also search NHS approved healthcare advice, and schedule GP appointments.
Business models based on partnerships
To settle into the market, apps like Ada have to fit into the complex healthcare ecosystem and constantly expand their functionality in accordance with patients’ expectations and market trends. In the UK Ada is partnering with Doctify (a platform for patients which lets them find a doctor and book a visit), Dr Julian (a therapist platform), DocTap (appointments with private GPs in London). Claire Novorol has bigger ambitions – she wants Ada to become a health concierge for patients, helping them to become more aware of the current state of health, strengthening prevention, monitoring health risks and goals. To make it happen, the way AI communicates with the patient must be simplified, the health assessment must be based on natural language.
Vincent Jörres (Ada Health) says that in the future Ada will continue to provide increasingly more sophisticated and personalized health assessments. She will be able to take into account the full medical history where available, as well as data from an even broader range of sensors, devices and other sources. It will strengthen the prevention approach. “Besides that, we are working on a “doctors’ version” to support physicians even more comprehensively in their daily work,” confirms Vincent Jörres.
Mobile apps must evolve and constantly add new features to ensure largescale use. Babylon recently launched Healthcheck, which enables users to understand how their physical and mental health may be affected by current and past lifestyle choices, as well as medical and family history. Healthcheck makes use of the information provided to build a condition risk profile and shows how each risk factor (e.g. alcohol intake, smoking) can statistically influence the chances of different diseases developing. Extensive epidemiological research has been used to quantify disease risk factors by demographic, and their effect on disease risk. Babylon is also embedded into Samsung Health, which is available on millions of Samsung mobile devices in the UK and US, and has recently signed major partnership agreements with Tencent, Bupa and Prudential.
Parallel to market growth through cooperation, the apps’ skills are being expanded towards seamless communication between human and the machine. “Symptomate will develop its conversational and natural language capabilities. We want to make sure that you can speak with the app in the same way you’d talk to a real doctor. Also, we plan to incorporate additional data points such as vital signs captured by wearables devices, lab test results and also elements of emotional intelligence,” answers Piotr Orzechowski from Infermedica when asked about future plans. Just like Ada, Infermedica’s strategy includes first of all partnerships with healthcare stakeholders instead of building a competitive advantage individually. “Our business model is B2B-focused. We work with health insurance companies, hospital systems and healthcare providers and provide them with highly customized and white-label solutions for patient triage and clinical decision support. Some of our key partnerships include leading European healthcare insurance companies such as Allianz, Dovera, Medis and PZU Zdrowie,” admits Infermedica’s CEO.
Symptom checking, mostly available for free for patients, is a complementary function to reach target groups. Patients are not ready yet to pay for AI-based services. To ensure revenue, companies like Ada, Infermedia, Babylon or Your.MD implement additional services like paid telemedicine consultations. Furthermore, they establish B2B partnerships with other healthcare stakeholders (insurance companies, private primary health providers).
From symptom checkers to smart health assistants
To make healthcare accessible, affordable, and available 24/7, including in rural areas, to empower patients, to improve quality of care, to strengthen the role of prevention or to democratize healthcare – mobile apps for diagnosis will play an important role on healthcare markets. Patients are getting more empowered and want to take control over their own health or get a pre-diagnosis before visiting a doctor. Also healthcare organizations must optimize their processes to ensure high quality of care, enhance patient experience, delivering the right diagnosis quickly and efficiently. It doesn’t appear that one leading solution will monopolize the market – it’s crucial to integrate with a local healthcare services architecture and set partnership to ensure the growth and market penetration.
Although these kinds of apps are based on a certain portion of data declared by a patient, they can’t see or hear the patient and perform a physical examination, they can act as a reliable source of medical information and deliver pre-diagnosis with a high accuracy. In the future they will have to evolve towards patient’s concierge, analyzing data from wearables and electronic medical records.
“Technologies like Ada will put the patient in the driver’s seat, allowing them to have more responsibility and control over their data and their health, a lot of it even from home incl. monitoring and managing chronic conditions, changing habits,” concludes Vincent Jörres from Ada Health.
Note from the author: Most of the mobile apps for symptom assessment are not allowed to “diagnose” – in the literal sense – patients. In many national health systems the “diagnosing” process requires face-to-face interaction between the doctor and the patient. In this case substitute terms like “symptom evaluation”, “preliminary health assessment”, “symptom checking”, “pre-diagnosing” are in use.
The 5 most popular symptom checkers on the market
- Infermedica – Symptomate, an online platform for symptom checking and clinical decision support
- Ada Health – free symptom checker (app, website), also available for voice assistants
- Your.MD – symptom checker and health tracker (app for diagnosis, partners with online medical service providers and refers its apps users to pharmacies, test centers, doctors’ offices or recommends other medical apps which are suitable for them)
- Babylon – free health service and health app with the main functionality of remote doctor consultation incl. a chatbot for diagnosis
- Sensely – an app for pre-diagnosis, video doctor consultation, allows remote monitoring, links user with local medical services and self-care resources. Currently the Sensely app is available via an access code from an employer/health plan.