AI for Preterm Babies
Infer the circadian rhythm of preterm babies from UMC patients monitoring data and develop ML models to regulate light and sound to improve their development in incubators.
Premium participation: Get 1-on-1 Mentoring & GuidanceTry a free mentoring session
Develop ML models to improve the development of preterm babies in incubators by optimizing their circadian rhythm with light and sound.
FruitPunch AI for Health has teamed up with UMC Utrecht once again for to improve the development of preterm babies in incubators. UMC Utrecht uses high tech incubators to extract data from preterm babies’ day/night cycles - light measurements, phototherapy, heartbeat frequency, etc. we can use this to determine the optimal circadian rhythm for baby development.
Deploying correlative action and detecting patterns of behaviour in the babies during different light and sound conditions will aim to yield a definition of rhythm, which is then tested for potential better development of the babies. To collect the data for the patients, the hospital used special cutting edge equipment that detects sound waves, measures light levels precisely, the pulse of the baby inside and many more factors!
Analyse preterm baby data, determine their circadian rhythm and how it can be utilized to improve a baby’s development.
Our challenge partner
Why we're helping preterm baby development in incubators with AI
“Ten percent of the babies in the Netherlands are born preterm” - this means a high risk of development issues related to brain growth, overall weight gain and the relative IQ score increase over time.
Scientists in the UMC Utrecht have dug into the circadian rhythm of babies in wombs, where they found a stunning correlation between the day/night cycle and the healthy development of a baby before it is born! Nowadays, technology allows for babies that are born earlier than expected to be put in devices called incubators that simulate the womb of a mother. However, it is not exactly clear how rhythm can be simulated within those incubators. Extensive data analysis and machine learning techniques are therefore in demand to understand how rhythm can be successfully simulated!
GOAL: Determine the optimal circadian rhythm for preterm babies in incubators from historic data, and develop a machine learning model to automate rhythm maintenance.
We will be developing data scientific, probabilistic and statistical methods to define what rhythm will be most effective when aiding babies in developing their bodies fully.
The subgoals include:
- Search for correlation between the light and audio factors and the growth/health of the babies.
- Use unsupervised models to determine how the factors are related to each other
- Use time series models to determine how the patients change over time.
To determine the definitions of a rhythm we will use the following:
- Probability Distributions
To evaluate each definition of rhythm, we will use:
- ROC and Confusion Matrix research
Who are we looking for?
For this Challenge we are specifically looking for students and young professionals located in the Netherlands. The Kick-off and first masterclass will take place at the UMC in Utrecht.
Furthermore, we are looking for data science, AI engineering + medical students and professionals within the Netherlands, preferably experienced in statistical inference, exploratory data analytics, visualization, machine learning model building, evaluation and explainable AI.
- AI engineer
- Data analyst
- Medical student
- Data science student
From the AI/Data Science/Computer Science contributors we ask for the input of 8-12 hours a week. Medical students will take up different tasks within the teams and will need to be accountable for ~4 hours a week.
We’ll organise 2 masterclasses on relevant topics during the challenge to bring you up to speed.
FACTS: Did you know?
🍼 1 in 10 babies born in the Netherlands is born preterm.
👶 During fetal life and long before birth, the mother entrains the developing circadian rhythm of the infant to the light-dark cycle.
🧸 Research has shown a delay of 2 to 3 weeks in development of melatonin circadian rhythm in preterm infants compared to full-term infants.
👼 The presence or absence of circadian rhythms in the infant after birth results from the combined influence of prenatal and postnatal environmental conditions.
Application Deadline: 8 January 2023
Challenge Kick-off: 16 January 2023
Midterm Presentation: 20 February 2023
Final Presentations: 6 April 2023