Lost in uncertainty

Artur Olesch
Jan 8 · 6 min read

Patients often have to wait years to get a correct diagnosis. Some of them are left forever misdiagnosed. In a desperate search for the truth, they become detectives investigating their own problems, going from one doctor to another and doing detailed research on the Internet. What to do when nobody knows what’s wrong with you?

Illustration by Tomek Tuz

Harmful information gaps

When you feel unwell or are concerned about your health, there is a natural urge to get a doctor’s opinion immediately. The first step to eliminate health problems is to get a correct diagnosis. Next, modern medicine steps in, with a range of treatment options and drugs. Sometimes a single pill can cure a condition; in other cases, advanced medical interventions are required. Regardless, the patient can count on the best possible help from professionals supported by advanced technical and pharmaceutical capabilities, provided the cause of the symptoms is known.

When medical examinations fail, the consequences may prove truly harmful. Misdiagnosis is more common than drug errors, although the scale of the problem remains unknown. A study published in the American Journal of Medicine suggests that up to 15 percent of all medical cases in developed countries are misdiagnosed. That means that one in every seven diagnoses is incorrect. This might only be the tip of the iceberg, as most health systems lack adequate or mandatory reporting. Delayed treatment may have damaging consequences, often lasts longer, and is usually not as effective as early intervention. Thus, the patient’s quality of life is affected. Apart from the harm done to patients, huge additional healthcare costs arise. And most worrying of all, estimates suggest that 1.5 million people worldwide die each year due to misdiagnosis.

Recognizing the hidden

A paradox of today’s medicine is that even though we are able to successfully treat more and more diseases, patients are not cured because the most vital part of healthcare often fails: the correct diagnosis. There are numerous reasons for this. First of all, some medical cases are not easy to recognize. Professionals are often reluctant to ask senior colleagues for a second opinion. They judge the patient’s symptoms too quickly, ignoring nuances. Some are biased towards certain individuals or are simply overworked, because they see too many patients throughout the day. What’s more, patients don’t always give a precise account of their symptoms. Stress and time limitations are not conducive to communication between a patient and doctor. Less often, misdiagnosis is the result of errors in laboratory tests or medical imaging.

The second set of reasons has its roots in education. There are 30,000 known diseases in the world, many of which have nonspecific symptoms. Among them, over 6,000 conditions are defined as rare. A disease is considered “rare” or “orphan” in Europe when it affects fewer than 1 in 2,000 people. In the United States, these terms apply to diseases affecting fewer than 200,000 people. According to the EURORDIS, a non-governmental alliance of rare disease patient organizations, 30 million people are living with a rare disease in Europe and 300 million worldwide (3.5–5.9% worldwide population). Fifty percent of them affect children. Rare diseases are characterized by a broad diversity of symptoms that can vary from patient to patient. Even symptoms common for the flu may hide underlying rare diseases.

Let’s face the facts: Statistically, a doctor deals with around 300 of the most widespread diseases. Nobody is able to remember the specifics of all existing 30,000 conditions. What’s more, doctors have to make a quick decision within an average 10-minute slot for one visit, often without the possibility of consulting with other professionals. Some of the more thorough tests are not provided by local laboratories. Reimbursement policies and procedural guidelines fail to address patients with rare diseases. Over time, as doctors master a narrow field of medicine and encounter patients with similar problems, the general medical knowledge acquired at university has a tendency to shrink.

On the one hand, this is a positive development — professionals can recognize diseases and plan treatment faster. Unfortunately, when a nonspecific medical case occurs, the danger of misdiagnosis rises. Traditional education won’t solve this problem. Each year 2.5 million new scientific papers are published, and the number climbs 8–9% a year. More than a million biomedical-related articles appear each year on PubMed, a search engine for peer-reviewed biomedical and life sciences literature. Yes, we are gaining more knowledge, but only a small percentage can be applied to clinical practice.

Making the best of data

According to the Rare Disease Impact Report, it takes 7.6 years in the USA and 5.6 years in the UK for patients with a rare disease to receive a correct diagnosis.

Every patient wants to get better. When a suggested course of treatment doesn’t work, they don’t give up. Undiagnosed patients look for help by visiting other doctors, often paying thousands for private consultations. They desperately Google symptoms. All of these have an enormous negative impact on a patient’s life. Everything changes when uncertainty rules your life.

To change this, a holistic approach to every patient is necessary. This would require employing communication and information technologies, including artificial intelligence. In the art of diagnosis, doctors’ experience and intuition should be supplemented with the ability of algorithms to analyze large data sets. Although doctors already have unlimited access to current medical knowledge, this access is only theoretical. Analog data processed by human beings leads to information overload. This is the place for AI and symptom monitors to step in. Algorithms can link even syndromes that don’t share any symptoms and find similar cases.

Illustration by Tomek Tuz

“Diagnostic and treatment AI-supported tools that integrate data from various sources are being developed in Europe and elsewhere. Further investment is required to scale-up these solutions, deploy them in care settings and evaluate their impact as well as the ethical and regulatory aspects,” says Virginie Bros-Facer, Scientific Director at the EURORDIS-Rare Diseases Europe.

Support from AI is part of a complex solution. Another essential factor is strengthening international cooperation. Knowledge from clinical research cannot remain isolated in data silos. This slows down progress in medical sciences, especially now that Big Data and AI have come into play. Sharing knowledge brings about benefits for all. The best example is European Reference Networks, a project that aims to tackle complex or rare diseases and conditions requiring a concentration of knowledge and resources. There are already 300 hospitals in 25 European countries exchanging data to provide more precise diagnoses. It appears that such partnerships actually work, as confirmed by a study conducted at the Stanford University School of Medicine. Doctors cooperating under the umbrella of the Undiagnosed Disease Network, a program created by the National Institutes of Health (USA), have diagnosed 100+ previously unsolved cases.

Besides AI support and networks, there is one more important factor: patient empowerment. Individuals are the best advocates for their own health, and they should also have access to diagnosis alternatives. In many cases, it’s the patient who finds the right diagnosis in the end. They don’t give up the search for answers, because the unknown disease affects not only their own life but the lives of their families as well. Unfortunately, they don’t have many options. Besides follow-up appointments and seeing specialists at major university hospitals or academic medical centers, these patients do extensive research on the Internet and sign up for patient groups. Recently they have also gained an ally in symptom checkers that ask detailed questions, check the answers in medical databases, and suggest possible causes.

Medicine is an art, but first and foremost a science. Solving the problem of undiagnosed patients is one of the most complex challenges faced by medicine. Combining science, a holistic approach, cooperation, patient empowerment, and new technologies such as artificial intelligence will lead us toward success. Evidence-based medicine and diagnosis require data. Let’s stop overlooking data and start taking into account even the smallest bits of information, which may be the missing pieces needed to complete the overall picture of a patient’s health.


Originally published at https://blog.infermedica.com on December 13, 2019.

Infermedica

Stories about AI-driven healthcare.

Artur Olesch

Written by

Digital Health Journalist

Infermedica

Stories about AI-driven healthcare.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade