The report “Artificial intelligence in the clinic” was written in collaboration with a broad-based expert group, and presents the possibilities for the new technology in the Norwegian health service.
“We can encounter artificial intelligence already from our very first contact with the health service, and then during diagnosis, treatment, follow-up and prevention. Among other things, we can receive faster and more personalised cancer treatment and better follow-up of those with chronic illnesses,” says Tore Tennøe, Director of the Norwegian Board of Technology.
“But we also face complicated questions relating to privacy and sensitive health data, who is responsible for medical decisions, and the risk of discrimination,” he continues.
The report reviews six trends for artificial intelligence in healthcare, and what these trends could mean for the Norwegian health service.
“The decisions that are now being made in the Norwegian parliament and in the clinics will be of decisive importance to what the services offered to patients will look like in ten years. We want to contribute to the debate on what the health service will look like in the future,” says project manager Anne Siri K. Bekkelund.
Artificial intelligence in the clinic – Six trends for the health service of the future
Six trends for artificial intelligence in healthcare:
1: The first line goes digital
Computer systems with artificial intelligence can speak the patient’s language and respond quickly and accurately when people make contact. Digital first lines can provide better healthcare across the country, and enable health personnel to spend less time on the telephone and more time on treatment.
Example:
At the start of the coronavirus pandemic, the emergency number was blocked by the sheer number of concerned callers. HelseNorge’s Koronasjekk (Corona Check) is a simple chatbot that has made information more easily accessible to people, and provided relief to health personnel.
2: Health personnel are assigned digital assistants
Digital health assistants can assist health personnel in making a diagnosis, selecting treatment, monitoring the patient and warning of complications. By using machine learning, they can analyse medical literature, interpret images and other data, and plow through thousands of patient records. This may become the key to treatment that is better adapted to each individual, and less dependent on the doctor’s experience.
Example:
At the hospital in Ålesund, they have implemented a system that can automatically draw the heart, lungs and other organs on a digital CT image. It was previously the case that specialists with extensive experience spent many hours on this work. Now they can concentrate on using the images to help the patient.
3: Diagnosis and treatment merge together
Artificial intelligence assists with the patient being assessed, receiving a diagnosis and being treated at one and the same doctor’s visit.
Example:
A new technique provides images of tissue samples of the brain almost instantly, and with the help of machine learning, the analysis time can be reduced from at least half an hour to two to three minutes. This means that surgeons can immediately operate and remove a malignant tumour in the brain and have the scull open for significantly less time, thereby reducing the risks associated with the operation.
4: Everyone can monitor their own health
Watches and wristbands with sensors can register everything from heart rate to tone of voice. Artificial intelligence interprets the data and provides continual information about the users’ physical and mental health. This makes it easier to monitor patients at home, detect illness earlier and start treatment more quickly.
Example:
Ten per cent of Norwegian municipalities used digital home monitoring during the coronavirus pandemic. In Larvik, they monitored patients in risk groups with digital thermometers and pulse oximeters. The readings were automatically monitored, and if the system detected anything serious, health personnel were contacted.
5: Equipment is constantly improving itself
By using artificial intelligence, the software in medical devices can continuously learn from all new data, and continually improve and update itself. The development can lead to faster improvements in healthcare because the physical equipment does not need to be replaced in order to access improvements. This can be particularly beneficial in health situations that change rapidly, such as during the coronavirus pandemic, or to detect patterns that change during seasons or from year-to-year.
Example:
An algorithm that uses heart rate data to detect abnormalities can be improved by making it a continuous learning algorithm. It can then better take into account seasonal variations, variations between patients or variations between clinics.
6: Prevention is tailored
By using machine learning, the health service can become better at finding people who are at greater risk of illness, and artificial intelligence can revolutionize the screening programmes. Machine learning also increases the possibilities for determining which preventive measures actually have an effect, and can be used to adapt recommendations to each individual.
Example:
Algorithms can select which people should be prioritized for breast cancer screening, thus increasing the probability of finding those who actually have cancer.
The expert group for the project:
- Helga M. Brøgger, Medical Doctor, Head of the Norwegian Radiological Association.
- Erik Fosse, Chief Attending Physician at the Intervention Centre ,Oslo University Hospital.
- Steinar Madsen, Medical Director at the Norwegian Medicines Agency.
- Damoun Nassehi, General Practitioner, researcher at the University of Stavanger and member of the Norwegian Board of Technology.
- Michael Riegler, Chief Research Scientist at SimulaMet and Associate Professor at The University of Tromsø – The Arctic University of Norway (UIT).
