
Philips uGrow: In-app Personalization Strategies
Context
Philips has developed the world’s first medical baby monitoring app - uGrow (rebranded as Baby+). The app functions as a logbook which allows mothers to track how they take care of their baby.
During my master thesis project, I joined the Data Science group at Philips Research in Eindhoven and worked on the uGrow project.
Team:
Individual supervised work
Mentored by Philips Research and Eindhoven University of Technology
Methods:
Questionnaire Design
SPSS
MPlus
The Challenge
uGrow hasn't fully made use of the digital data that it has access to, so it merely works as a tracking app.
Based on my literature review, I believed that mothers who are parenting infants would benefit greatly from some assistance in order to be a competent caregiver.
It requires a solution which not only provides care-taking knowledge, but also helps mothers to optimize their psychological states, e.g. providing suggestions on how to cope with parenting stress.
Research
Infant parenting has been extensively examined by researchers from different academic fields and they proposed the Infant Parenting Model (see below).

Based on this model, my research study aimed to answer two questions:
-
What parenting beliefs and behaviours should be focused on to provide personalization?
-
What type of personalization service can be provided?
Research Questions
In order to achieve the goal above, three research questions need to be answered:
RQ1: What are the parenting beliefs that can influence parenting behaviours? How are these beliefs related? And how are they related to parenting behaviours?

RQ2: What role do other determinants play in the relations mentioned in RQ1?

RQ3: Can we describe a mother in terms of parenting beliefs, parenting stress, or perception of infant difficulty, based solely on the app-tracked parenting behaviours?

Questionnaire Design, Recruitment and Data Analysis
In order to answer my research questions, I designed the online questionnaire in adoption of different (sub)scales from different existing questionnaires.
The questionnaire includes six sections:
-
Demographics
-
Parenting beliefs in structure and attunement
-
Maternal self-efficacy
-
Parenting stress
-
Perceived infant difficulty
-
Parenting practices (based on the functionalities of uGrow).
The questionnaire is focusing on mothers who have an infant(s) who’s less than 1 year old.
Participants were recruited from UK (online) and the Netherlands.

Questionnaire example
I conducted Exploratory Factor Analysis (EFA) in SPSS and Structural Equation Modeling (SEM) technique in Mplus with a sample size of 202.
Key Insights
-
Mothers show a matching pattern between their beliefs and behaviours.


-
The matching pattern is not perfect, due to the influence of other variables.

-
A vicious circle was also found: The more difficult the baby is perceived to be, the less confident the mom feels, which causes the mom to be more stressed and then perceive the baby to be more difficult.
