Clinicians and researchers at St. Olavs Hospital and NTNU are able to detect brain injuries and Cerebral Palsy early through observation of infants' spontaneous movements. In the "In-Motion" project, video recordings are sent to experts with specialized experience in clinical examinations. In addition, machine learning algorithms and models for automatic tracking of Cerebral Palsy are being developed. CheckWare plays a central role in the transmission of the videos.
Around 120 babies and toddlers are affected by Cerebral Palsy (CP) annually in Norway. This often concerns premature infants, children who have had insufficient oxygen during birth, or children who have suffered brain hemorrhage.
Some children with CP are identified during follow-up at the hospital, others in primary healthcare. Since it is easier to observe the functional impairment when the children start crawling, standing up, and walking, most affected children receive the diagnosis between one and two years of age.
- Cerebral palsy is a condition that cannot be cured. There is no medication for the disease. Those affected have reduced motor function to varying degrees, some milder, others more severe. We want to influence how much the functional impairment actually becomes and aim to slow the disease through prevention, says Lars Adde. He is responsible for all medical research activities in the In-Motion concept at St. Olavs Hospital and NTNU.
The solution is a win-win-win situation both for children with the disease, children who are ruled out for the disease, and for the healthcare system. For patients with CP, it is very important that the disease is detected early so they can receive early assistance. At the same time, the value of ruling out CP is great for concerned parents. Not least, it will impact healthcare resources by confirming or excluding the diagnosis early.
The earlier, the better
Adde has more than 22 years of experience working with diagnostics and treatment in the neonatal intensive care unit, as well as follow-up programs for sick newborns in specialized healthcare. He has a background as a physiotherapist specializing in pediatric physiotherapy, as well as a PhD in clinical medicine.
He cannot emphasize enough how important it is to assess newborns early enough. With advanced CP, there is a risk of other complications such as joint stiffness, pain, nutritional difficulties, reduced vision and hearing, which in turn can lead to repeated hospital admissions and surgeries.
It is important to identify who may become ill early, but it is equally important to find out who does not have or will develop CP. That 8-10 percent of children in the medical risk group get CP also means that about 90 percent do not get CP.
- The solution is a win-win-win situation both for children with the disease, children who are ruled out for the disease, and for the healthcare system. For patients with CP, it is very important that the disease is detected early so they can receive early assistance. At the same time, the value of ruling out CP is great for concerned parents. Not least, it will impact healthcare resources by confirming or excluding the diagnosis early.
Babies are filmed at home in normal, familiar surroundings
The clinical observation method for analyzing the child's movements is called General Movement Assessment (GMA) and is performed when the child is 2-4 months old after the original due date for birth. The method is recommended to identify children who are at high risk of developing CP. It can then be quite certain whether the children are at high or low risk of developing CP.
GMA was developed in the early 2000s and has been extensively scientifically documented to this day. The child is filmed for three minutes while lying on their back and moving undisturbed so spontaneous movements can be observed. Afterwards, the video can be analyzed using GMA. Trained clinical observers perform the assessments.
- For many years, the check-ups were only conducted at the hospital. To carry it out, parents and the child had to come in for an appointment. Many things had to fall into place at the scheduled time: The child had to be awake, active, comfortable, and generally well when the recording was taken. Now, the recording is done at home when it suits the child best. At ease, the parents get the opportunity to film without having to travel to the hospital, explains Lars Adde.
He adds: - This not only contributes to better quality of the recording but also makes the parents more involved in the process, instead of being passive observers.
The videos are sent from the parents to the hospital. The GMA assessment is done remotely before the family is informed about the next steps, including follow-ups. This makes it possible to detect signs of CP at an earlier stage.
For many years, the check-ups were only conducted at the hospital. To carry it out, parents and the child had to come in for an appointment. Many things had to fall into place at the scheduled time: The child had to be awake, active, comfortable, and generally well when the recording was taken. Now, the recording is done at home when it suits the child best. At ease, the parents get the opportunity to film without having to travel to the hospital. This not only contributes to better quality of the recording but also makes the parents more involved in the process, instead of being passive observers.
Incidence of Cerebral Palsy in Norway
Cerebral palsy (CP) is the most common cause of motor disability in children and occurs in approximately two per 1000 live births (FHI.no)
120 Norwegian children are diagnosed with Cerebral Palsy each year (FHI.no)
There is no cure for CP, but treatment can relieve symptoms and improve motor skills (FHI.no)
CheckWare with video solution for home use
For the In-Motion project, it has been crucial to choose the right solution for remote treatment, especially considering privacy. There are challenges related to, for example, using smartphones with video recordings of the patients, so the project had to explore the market for setups that complied with all privacy regulations.
- We have worked for a long time with different solutions, including an app that was very expensive. Eventually, we came across CheckWare, which had a system for sending questionnaires to patients. The solution was regulatory approved by Helse Midt-Norge, which was important for us, explains Adde.
HEMIT (Helse Midt-Norge IT), the IT organization at St. Olavs hospital, was tasked with developing a customized solution for submitting video files from parents with premature babies. A system was set up in collaboration with CheckWare: This included, among other things, automatic messages via SMS to parents, and the ability to receive training materials digitally.
- The involved parents use the CheckWare solution to send the videos to us. We get a unique overview to see who has received messages, submitted videos, received reminders, and so forth. For us who collect the videos, this is a revolution. Where we previously worked with Excel sheets and manual operations via phones and Outlook, the CheckWare solution today is fully automated, says Adde.
He emphasizes that they can now confidently carry out good quality control. Through surveys and questionnaires, they ensure that parents find the method to be a safe way to film and transfer videos of their children.
- Parents think this is a really great way to submit videos. It feels secure and feasible, and just as safe as having to go to the hospital to film.
- I highly recommend the CheckWare solution, even though it was only launched six months ago, he says, highlighting a key person at HEMIT:
- Per-Henning Valderhaug is an advisor at HEMIT. He has done a great job getting the solution in place and has been the link between the clinical environment and CheckWare. He definitely deserves recognition because he has been crucial to the project, Adde believes.
The involved parents use the CheckWare solution to send the videos to us. We get a unique overview to see who has received messages, submitted videos, received reminders, and so forth. For us who collect the videos, this is a revolution. Where we previously worked with Excel sheets and manual operations via phones and Outlook, the CheckWare solution today is fully automated.
Machine vs human
Behind In-Motion is a multidisciplinary research team consisting of physiotherapists, neonatologists, pediatricians, movement scientists, and computer engineers. The In-Motion team covers clinical national and international studies at several sites in Norway, Denmark, the USA, Belgium, China, India, and Turkey, together forming a database for multicenter studies.
Adde is a specialist in pediatric physiotherapy, GMA observer, and knows everything about infants’ spontaneous movements. On his team he has Espen A. F. Ihlen, responsible for the machine learning algorithm together with Daniel Groos at NTNU. The latter defended his thesis just a few weeks ago. A critical expert environment behind them is the Norwegian Open AI-Lab, headed by Professor Heri Ramampiaro at NTNU. This environment will play a central role in the continued development work with automation and digitalization in In-Motion.
It is worth noting that In-Motion is based on two "pillars": One pillar uses medical expertise to classify spontaneous movements and assess the risk of developing CP. The other pillar uses trained machine learning models to do the same with automatic image analysis. Both pillars use 3-minute-long video recordings of the infant’s spontaneous movements at 2-4 months of age as a basis.
- We train a machine learning algorithm to recognize patterns in the video images. Then we use this to indicate the likelihood that the child will develop CP. The automated results are meant only as decision support in the decision processes, which are still conducted by human experts, says Adde.
He explains that the processes are quite complicated and that the expert’s brain must be trained to recognize movement patterns in a baby. The competence should be maintained, while they also work on recruiting new experts.
The dream of automated movement analysis
In-Motion currently follows 50-60 children annually in Helse Midt-Norge. On a national basis, they envision 700-800 children per year.
- I have had this dream of video recording and automatic movement analysis for 20 years. Remote early risk assessment for CP development in all sick newborns in Norway is now actually within reach. This is thanks to finding a solution where we could receive videos in a safe and patient-secure way, he says, explaining that the video solutions are so far only available in Helse Midt-Norge.
Lars Adde is part of a service innovation project in Helse Midt-Norge together with Beate Horsberg Eriksen at Ålesund Hospital, Kristin Åberg at Levanger Hospital, and Kristine Grünewaldt at St. Olavs hospital.
- We are now working to implement the solution in more places in Norway. We want to expand the CheckWare solution to other health regions for remote expert assessment.
Adde adds: - The ultimate dream is for the whole world to have access. But first, we need to test a pilot with the algorithm. Initially, it is realistic to work towards hospitals throughout Norway. We must ensure that as many new families as possible have access to home examinations as early as possible.
