Artificial Intelligence in healthcare still seems like a fairy tale in many Latin American countries.
However, companies like Arkangel AI are helping the healthcare industry use algorithms to diagnose and treat diseases without having to write a single line of code.
"What we do is create technology to help health professionals or health systems deliver results much faster and much more scalable without any knowledge of coding or artificial intelligence necessary," said Laura Velásquez, President of Arkangel AI, in an interview for Omnia Health.
Velásquez explained that the company is helping with the early diagnosis and effective treatment of different diseases with the application of AI.
A diagnosis of fibrosis can typically take weeks for example, in which dozens of studies are reviewed by specialists who analye the case of the patient.
The same occurs with lung cancer, which can take up to three months to be correctly diagnosed; when this happens the patient is already at a late stage and there is not much that can be done.
With algorithms developed from AI, the verification of a medical condition can be reduced to two minutes and help the doctor make better decisions on medical treatment.
In addition, it is 90% accurate, as required by the WHO in the use of these technologies.
Velásquez assured that algorithms can also help doctors to enhance treatment, as with COVID-19: Arkangel developed a prognosis model for the patient with COVID when they were already in a clinical state.
When the pandemic began, she recalled, it took up to 36 hours for PCR test results to arrive and doctors did not know what to do to treat the patient, while confirming the disease.
“We decided to release these algorithms for free and help different municipalities in Colombia,“ Velásquez explained. “They used Arkangel to say ‘hey this is happening to the patient’s lung and we can do this, we can do that, it may be COVID or it may be something different.”
How Arkangel’s algorithms work
Arkangel has algorithms already developed for respiratory diseases, retinal diseases, parasitic diseases, such as malaria and Chagas disease, or bacterial, delivered to doctors, pharmaceutical companies or government health institutions the algorithms already trained for diagnoses or treatments.
“They acquire our license and the rollout begins. Our licenses are five times more affordable than those that currently exist because it’s 100% software and that allows us to make it more accessible,” explained the president of the company.
The second option available is for hospitals, companies or government to enter all data available into Hippocrates, software that analyses the data and generates a model that is then trained by the algorithm for a specific purpose.
“In the end, after several hours of training that can be from 40 to 200 hours, depending on the complexity of the data and the pathology, we deliver the model to you. This is an algorithm that enables early detection of diseases, or triages the patient, or whatever the entity requires, ”said Velásquez.
The advantage of using these algorithms and computers is that they do not tire and therefore they maintain their degree of precision, compared to a doctor who sees 30 patients a day, and who after eight patients already has the normal exhaustion of any person.
Furthermore, health centres, clinics and hospitals do not need large infrastructure to take advantage of this technology. Data from health systems show that 92% of hospitals have at least one X-ray machine.
"Although almost all providers are analogue in rural areas, we can use images still, because the doctors take a photo with their cell phone, they put it in Arkangel and they get the result," said Velásquez.
In the more remote areas where there is no internet, models can be worked on offline without any loss of precision, while remaining as fast as possible for the doctor.
Accessible healthcare, thanks to algorithms
Velásquez nonetheless assured that her dream was to take algorithms already trained in large urban hospitals to more remote communities, where there are no specialist doctors and where access to health services is more difficult.
"That is what we want - we have algorithms that we can already use in the field, in urban areas, in rural areas, whatever, and with these new ones we can train with Hippocrates say, 'Why can't we take it to any area that doesn't have access to this - it could be in Latin America or Africa or wherever',” she said.
Laura is confident that the industry will continue to embrace these new technologies and implement them in more healthcare systems and hospitals.
The next challenge, she says, will be the selection of patient data to continue training algorithms since, until now, no one has been given the task of classifying this data for use in personalised patient medicine.
Hospitals will further have better patient management and doctors will be focused on patient care, without having to learn to code these algorithms.
“We allow you to continue doing what you do, which is to be doctors, to provide patient care, and we take care of the back end that nobody sees as it is encoded around AI,” Velásquez concluded.