Mental health algorithms with the ability to mimic empathy? WIll A.I. be smarter than human doctors? Simple big data analytical software presented with clever marketing tactics? It is hard to assess the actual situation when it comes to artificial intelligence’s involvement in healthcare. Moreover, there is no database containing all the smart algorithms that are worth applying to medical processes. Every artificial intelligence-based algorithm already approved by the FDA has been collected for the analysis – algorithms that are proven, reliable, and accurate solutions enabled by an official regulator for medical use. Let’s examine the infographic in detail!
Algorithmic healing and it’s factors
When aiming to assess the smart algorithms in healthcare, we took into consideration temporal and spacial factors, accuracy and credibility, as well as medical specialties where A.I. algorithms are able to better the care process.
In the infographic you can see that there is a noticeable uptake in the appearance of new solutions in the last few years, in regards to the timeline. AliveCor’s algorithm, able to detect atrial fibrillation was the only algorithm approved in 2014. In 2016, the FDA approved another four solutions prepared for clinical use and in 2017, a total of six new algorithms were also approved by the US regulator. Rapid growth appeared in 2018, when the FDA endorsed a total of 23 algorithms in the medicine field. The first approvals in early 2019, showed that the trend is not expected to slow down, on the contrary, we will most probably see multiple new medical A.I. solutions entering the market.
Regarding the spatial factors, the most crucial hubs for A.I. development are: Silicon Valley, the Boston-New York area, Montéal, London, Beijing and Bangalore. The same can be applied in medicine and healthcare. When compiling FDA approved algorithms, the most decisive factor is that it is the only criteria available for credible and accurate medical software. In Europe there is the European Medicine Agency, that includes guidelines and statements about Artificial Intelligence, however, the FDA is the only regulator that has efficient instruments in its toolkit able to meticulously assess the credibility and accuracy of algorithms for medical purposes. This factor also reflects why our information revolves only around the U.S. market and the developments considered are within the FDA’s jurisdiction.
FDA Approval. What does it mean?
Regarding the meaning of the FDA approval itself, the listed algorithms embrace the entire scale of approvals starting from 510(K) submission through de novo to premarket approval (PMA). The former concerns a premarket submission to show that a device aiming for market launch, (but not requiring premarket approval), is as safe and effective as other similar instruments with PMA. The latter concerns the FDA process of scientific and regulatory review to assess the safety and effectiveness of medical devices supporting and/or sustaining human life and the most stringent of the device marketing applications.
To address novel devices of low to moderate risk that do not have a valid predicate device, the de novo pathway for device marketing rights was added – for example in the case of software solutions such as smart algorithms. After a successful review of a de novo submission, the FDA produces a classification for the instrument, if needed, it also creates a regulation and identifies any special controls required for future premarket submissions of significantly equivalent devices. If the FDA approval had been separated into subtypes the infographic would have been too complex so for now it as a single category.
Medical specialties that are more algorithm-friendly
When studying the infographic, the distribution of smart algorithms in the various medical specialties is apparent. Radiology and cardiology are the two fields which have the most Artifical Intelligence-based solutions. In cardiology there are 7 approved algorithms and 16 in radiology. However, the fields of geriatrics, orthopedics and pathology seem to be less prone to A.I. Some other fields, such as pulmonology, dermatology, surgery, OB/Gyn or forensic medicine, are not even on the list yet.
Definitive conclusions should not be drawn only from this infographic however, as it is just a mere snapshot of the current situation and does not make any revelations about the trends. In the field of pathology, A.I. is a promising technology, but the number of algorithms that have been approved by the FDA are low for the time being. Pathology might need the next few years to catch up to the number of solutions already existing in radiology or cardiology.
There are several reasons why these two fields are at the top of the list when it comes to A.I. research. Firstly, computer vision is one of the fields with the fastest growth in A.I. development, and medical imaging has the data and the visuality that smart algorithms require to thrive. Consequently, commerical software for automatically classifying breast density and in turn, detecting breast cancer, are able to perform on par with human radiologists, researchers found. The FDA approved the first AI system that can be utilized for medical diagnosis, without input from a human clinician, in April 2018.
FDA approved algorithms in medicine
- AliveCor supports the early detection of atrial fibrillation, developed an ECG analytics platform – just as PhysiQ Heart Rhythm Module, Apple and Cardiologs – and a six-lead smartphone ECG.
- QbCheck aids the diagnosis and treatment of ADHD.
- InPen is used to track insulin dosage.
- One Drop Blood Glucose measures blood glucose levels and automatically sends the data to the paired app.
- Lumify provides ultrasound image diagnosis.
- Cantab Mobile acts as a tool for memory problem assessment for the elderly.
- EnsoSleep created a tool for recognizing sleep disorders.
- AmCAD-US evaluates thyroid nodules and separates the nodule characteristics into categories.
- Lepu Medical and BioFlux detect arrhythimas.
- Subtle Medical offers a medical imaging platform.
- Bay Labs provide echocardiogram analysis.
- Viz.AI detects stroke on CT scans and assists clinicians in winning the race against time.
- Arterys’ algorithm can spot cancerous lesions in the liver and the lungs on CT and MR images.
- Empatica helps to detect epileptic seizures.
- Cognoa’s algorithm built into an app helps diagnose autism in children.
- POGO and Medtronic monitor and predict blood glucose fluctuations.
- Idx autonomously detects diabetic retinopathy with the use of retinal images.
- Icometrix assists neurologists in interpeting brain MR images.
- Imagen helps healthcare providers to identify wrist fractures with accuracy rates similar to human radiologists.
- NeuralBot offers a solution for transcranial Doppler probe positioning.
- MindMotion Go advances its algorithm for motion capture for the elderly.
- Dreamed helps healthcare professionals in the management of Type 1 diabetes.
- Zebra Medical Vision detects and measures coronary artery calcification and analyses chest X-rays.
- Aidoc can flag brain bleeding in head CT images and pulmonary embolism.
- iCAD categorizes breast density and detects breast cancer as accurately as radiologists.
- ScreenPoint Medical assists radiologists with the reading of screening mammograms.
- Briefcase triages and diagnoses time-sensitive patients.
- RightEye Vision System tracks eye movements for indentifying visual tracking impairment.
- MaxQ develops an acute intracranial hemorrhage triage algorithm.
- ProFound AI detects and diagnoses suspicious lesions.
- ReSET-O offers an adjuvant treatment of substance abuse disorder.
- Verily developed an ECG feature on the Study Watch.
- Paige.AI offers a clinical-grade algorithm in pathology.
- FerriSmart created a machine learning solution for the measurement of liver iron concentration.