The ÜMA developments propose to expand telemedicine in Spanish-speaking countries with the assistance of artificial intelligence algorithms. “Health is in the midst of a revolution and technology has much to give to optimize the efficiency and quality of healthcare processes,” say its managers.
By Matías A. Loewy
“The health market is lagging compared to other markets, such as banking or commerce. And if you look at it in perspective, the maturation stage in the implementation of new technologies also had to face resistance and prejudice: ‘How am I going to trust making a transfer over the internet? Who guarantees me that the data transfer is secure? ‘ But the emergence of new technologies moves the status quo of the market, designs a different business model, to which not all established players adapt, and determines the emergence of new companies. We are one of them”, dice Jorge Estévez Rivas, an oceanographer, computer graduate and master in software engineering Spanish who, after making a career in the financial sector in London and working in Big Data in telecommunications, plunged into the sea of health.
Estévez, is the director of Engineering and Applied Artificial Intelligence of ÜMA, a spin-off of the company Emergencias (leader in the Argentine market for out-of-hospital emergencies) that seeks to evolve and accelerate the way of providing health services, in addition to enhancing them with artificial intelligence. It already has a presence in five Latin American countries and is also seeking to expand to Spain and the United States, where there are almost 60 million inhabitants of Latin origin.
Leading an international and multidisciplinary team of 24 people, which includes software engineers, doctors, psychologists and sociologists, Estévez affirms that they entered a “high performance team dynamic”. The vision: go beyond providing tools for teleconsultation and redefine health services, with an optics focused on patient access and using artificial intelligence tools to reduce errors and streamline processes.
For its part, Gustavo Daquarti, a cardiologist and DataScience specialist who serves as the medical director of ÜMA, ensures that “today, the health system distances the doctor from the patient. People have to wait a month and a half to access an appointment and the doctor, in turn, perhaps spends more time filling out forms than evaluating the patient. Our expectation is to transform this situation and, through technology, favor more accurate diagnoses and treatments ”.
More than 350,000 video consultations
The first development of ÜMA was a tool for online medical consultation, which, since the pandemic began in March, has made more than 350,000 video consultations, with a very high level of satisfaction (perceived quality) and with 60% of effective resolutions in this way.
“The initial aspiration was develop the best tool to meet the demand for low-risk consultations that do not need referral, and that many times involved sending a doctor to your home to solve a query as simple as a sore throat. And that he spent more time going from one place to another than treating patients, who could have the solution before the professional had to move ”, Daquarti points out.
When patients are treated in this way, they complete a questionnaire in which they indicate the reason for consultation, refer symptoms and even upload images (for example, of a skin lesion) through the smartphone, so that when the The doctor starts the virtual interaction and has already completed a good part of the anamnesis or questioning.
The tool is versatile, agile and friendly. The platform ÜMA It allows the patient to choose general practitioners or specialists according to the ranking made based on user ratings and comments.
ÜMA reaches its end users as a PWA (progressive web App) allowing a quick interface on any mobile operating system and quick integrations with other platforms.
The main clients of ÜMA in Latin America are Rappi (in Chile, Uruguay, Ecuador, Peru, Costa Rica and Argentina), Prudential, Klimber, UNICEF, YPF, Falabella and prepaid medical and state health institutions in Argentina and Ecuador, as well as partners in Costa Rica, Uruguay and Paraguay.
Prudential Insurance, for example, offers policyholders throughout Argentina the possibility of accessing teleconsultations through the ÜMA platform. And for members of the Vitality program, it also empowers them to serve ÜMA Wellness, which is a pack of specialties aimed at improving the quality of life where the user is.
“In times like these, it is a priority for us to be ever closer to our customers, providing safe and reliable tools that can help and accompany them in this context. We believe that through ÜMA, a means by which you can make unlimited virtual medical consultations for free, we collaborate with your health care always looking for the best way to protect them ”, highlights Gabriela Bordato, Business Development Manager – Individual Life at Prudential Seguros.
Solutions for epicrisis and diagnosis
But actually, online consultation is just one of the microservices of a broader platform named ÜMA Health, and that also includes a pioneering artificial intelligence algorithm in Latin America, Autonomous, which allows a assisted self-diagnosis of epicrisis (clinical summary at discharge or at the end of the consultation) that include the symptoms of COVID-19.
Autonomous is based on machine learning (deep learning) applied to natural language processing, explains Daquarti. “If there is something that doctors do not like is writing epicrisis, we all dream of one that is written automatically because we feel it as something bureaucratic or tedious,” he says.
The platform ÜMA offers a solution that is like an assistant who writes the epicrisis as the query is made. But that, instead of transcribing the dialogue as an automaton, it “decodes” that coming and going of questions and answers, transforms colloquial expressions in medical terms (for example, “headache” into “headache”) and then “interprets and he summarizes that information as if he were a doctor, ”says Daquarti.
Autonomous was already used in 170,000 of the video consultations and, in a next stage, the team of ÜMA proposes incorporate “more elaborate and intelligent” algorithms that, from a few sentences, they can continue writing the text and generate less structured epicrisis.
The tool also has the ability to use machine learning for the Autonomous predictive medical diagnosis in Spanish language, such as presented Estévez and Daquarti in the virtual congress Machine Learning for Health Care 2020, last August.
“These new algorithms allow the possibility of always having a colleague to consult what he would do in a certain situation. Since through this text structure, an algorithm trained with 300,000 medical records can be asked what is the most likely behavior that these colleagues would have followed in a similar situation. It is like a form of consultation, an endorsement of 300,000 visits for the diagnosis and the prescription of the prescription ”, says Daquarti.
The advance has implications for patient safety. “Medicine is one of the professions with the most errors, due to the complexity of the organism and nuances that the brain or the human eye cannot perceive. More than 35% of prescriptions are wrong. We are not far from that, with the help of technology, it is relatively easy to do better ”, emphasizes Estévez.
The medicine that comes
“The prediction of the diagnosis with the algorithm from natural language was shown to have an accuracy of 93%”, says Estévez, who predicts an explosion of complementary machine learning tools that, from audio or video signals, can determine the patient’s clinical parameters or vital signs remotely or with the smartphone.
“One of the applications is based on artificial vision (computer vision): the images are translated into numbers and these numbers are refined by a descending gradient until, for example, fever or heart rate is predicted ”, Estévez describes.
In the eyes of the specialist, the information technologies applied to health went through a stage of digitization of the processes and are now in a phase of helping the doctor and the institutions in the diagnosis and indication of the treatment, as well as in the monitoring in real time and monitoring of patients with chronic diseases.
“In 2 or 3 years, we will be closer to autonomous diagnosis, either with natural language or with biomarkers. Technology can help redefine processes and, for example, dispense with certain laboratory tests because they are unnecessary for diagnosis. At some point, there will be algorithms legally enabled to issue prescriptions at a level higher than the average for a human “predicts Estévez.
Whenever disruptive solutions emerge, a successive process of denial, resistance, boycott and acceptance takes place. “It is a natural process,” says Estévez. But the commitment of ÜMA it is not negotiated. “Universality is one of our dreams: to allow all people to access better health coverage,” he says.
More information: https://uma-health.com/