Op-ed: The livesaving potential for AI-enabled automated diagnosis of health conditions

By Jalil Allabadi

How our world is changing – automatic symptom identification will save lives  

As the digital revolution of the healthcare industry continues to take place across the globe – spurred on by the COVID-19 pandemic and the ongoing rapid adoption of technology - telehealth services are increasing access to healthcare for millions, without them having to step foot in a doctor's surgery.

While HealthTech, as it is now being called, can bring great advances in medicine through new cures and their applications, all patient diagnoses must first go through analysis of individual symptoms to determine the treatment required. This necessity has created a greater demand for health professionals, and their precious resource of time.

Today, advanced technology is facilitating the diagnosis of diseases, and patients no longer have to physically see a doctor to get an accurate diagnosis for their symptoms. And surprisingly, digital diagnostics can be even more accurate than those made by a human doctor.

Getting it right

Accurate and timely diagnosis at an early stage of a disease can have a significant impact on a patient's life, and sometimes mean the difference between life and death. However, there are several elements at play that can affect accurate diagnosis.

Firstly, the communication of symptoms is a key factor. The patient’s explanation of symptoms is usually made with an inherent heterogeneity and ambiguity, using unstructured and informal expressions. This can present difficulties for the doctor in interpreting which symptoms are present, as well as identifying and categorising them accurately against formal disease classification codes.

As is often the case, human error is another, with potential oversights made by (usually more inexperienced) physicians when identifying symptoms and determining the correct diagnosis. This is in part due to the massive number of symptoms that are common across different diseases, which can make accurate diagnosis unclear, as well as time-consuming.

To cope with growing demand clinical professionals are moving towards more telediagnosis. Thankfully, digitalisation can now offer the means to aggregate data and use newly developed programmes or apps to decipher the information gathered to assist doctors in reaching the right conclusions. This is an important development when it has been found that documenting patient data can take up 50% of doctors’ time.

The arrival of digital diagnostics is nothing short of revolutionary when the scale of the challenge is considered.

AI – doctors' new best friend

The world has been witnessing the evolution of artificial intelligence (AI) flourishing and influencing widely various real-life domains, including doctors' clinics.

An impressive advantage of online healthcare platforms is the large volume of data generated at different scales of variability. Analyzing this rich unstructured data has direct benefits in improving and personalizing diagnosis. Recent research studies have demonstrated the potential of AI techniques in disclosing possible symptoms contained in text. Essentially, recognizing the symptoms from text correctly results in a higher probability of determining the correct diagnosis.

But this can present challenges for complex languages, such as Arabic - the official language of over 300 million individuals across 22 countries the Middle East and North Africa (MENA) region. This comes in two forms; modern standard Arabic (MSA) used in formal situations, and colloquial Arabic, where it differs from one country to another and from city to city. Arabic has more alphabet characters and sounds that do not exist in other languages, with Arabic dialects ensuring the same word can be written differently (aside from misspellings too). This makes Arabic pre-processing a harder process.

Recently, machine and deep learning methods have enriched the diagnosis procedure by the automation of various processes. Now, automation through alphanumeric codes can map diseases into generic categories, including symptoms, standardizing the processing, presentation, and transfer of individuals' medical information via healthcare facilities – even in Arabic.

The resulting computer-aided diagnosis (CAD) systems are emerging computational tools that have demonstrated an ability to not just improve the reliability of decisions, but also, reduce the cost of patient monitoring, save doctors’ time, and provide patients with a better quality of care. Because so much of doctors’ time is spent recording data learned from patients, CAD results in improved productivity and better outcomes.

A study undertaken by medical researchers for Altibbi, an Arabic telehealth platform with over three million digital consultations stored on its database, points to the potential for significant improvement in patient diagnosis through automated symptom identification.

The findings revealed the AI model developed for diagnosis obtained a successful ability to predict at least one of the symptoms correctly from the patients’ descriptions.

Given the purpose of automated diagnosis is not to replace doctors but to assist them by providing additional tools to help sift information, the application of advanced technologies such as AI and deployment of new CAD systems is already confirming improvements in both accuracy and timeliness of diagnoses, which we already know saves lives.

It is a new healthcare reality that makes sense.
 

Jalil Allabadi is co-founder/CEO of Altibbi, the MENA's largest digital health platform – with over 20 million monthly users across 14 countries.

The views and opinions expressed in this article are those of the author and do not necessarily reflect the editorial policy or position of ArabMedicare.com.

PRINT THIS ARTICLE