When a couple’s baby is born prematurely, they find themselves in a situation for which no one is prepared. Every new parent starts with the same level of knowledge of premature babies and their care: little to nothing. The few months spent in a NICU is physically, emotionally, and mentally exhausting to the parents.

Nearly 15 million babies are born premature every year around the world. The percentage of prematures can reach 18% in certain parts of Africa and Asia, while in Scandinavia and other European countries it can be as low as 5-8%. The global average is 11%. Behind these numbers are not only the babies’ lives, but also those of their parents.

Small actions can make huge improvements. As a recent WHO report inspiringly suggests, mobilizing “parent power” in the care is a must. Innovative programs (Close Collaboration with Parents, Turku University Hospital, Finland; Family-Integrated Care, Mount Sinai Hospital, Toronto), have demonstrated improved outcomes with increased parental engagement.

Inspired by our lived experience with an extreme low-weight premature baby who spent nearly 5 months in two NICUs (UC San Diego and Stanford), we developed lumila, a parental engagement tool. In a larger design study with 300 parents we also validated the concept and identified other parents’ needs.

The product is currently offered to parents only. Next we will release the HIPAA-compliant version to involve NICU care teams. Taking families to the shared decision making lowers anxiety, increases helpfulness, and reduces long-term inpatient days.

And we can do much more with the power of AI. Our current efforts focus on users and usage scenarios: parental and collective users, as well as interaction and knowledge access:

1) conversational interaction: one finding of the design study was the need for seamless interaction. Using sensor technologies and na

tural language chatbots, the tool is able to contact the parent when left the hospital and ask for a daily activity summary and for emotional notes.

2) every-day language summaries: using Natural Language Generation, easy to understand recaps of the babies’ development status brings piece of mind and reassurance to parents during the day.

3) early warnings: by analyzing the correlation of the parental engagement and the babies’ developmental data, the care team can raise alerts and provide additional support to the parents for better engagement. This Deep Learn

ing-based AI component is currently work in progress.

4) shared decision making: the use of aggregated developmental and parental engagement data provides an opportunity for more informed decisions. We developed a visual interface, Duetto, that parents and doctors can jointly use while discussing the potential treatment options.

These scenarios form our long term vision of how lumila can be developed incrementally to a “Computational Care” solution, providing Patient Engagement and Shared Decision Making support. As healthcare is changing, patients need to become more and more informed and involved in their treatment.

That’s what lumila wants to enable. For now, our primary focus is supporting these tiny fighters and their parents.



Author: Peter Boda

Coauthor(s): We are a small startup in stealth mode. More information: http://lumila.us Our work falls into the Decision Support, Cloud & Big Data & Virtual Assistant topics, I chose the first category. Please re-group us in case you see that our work fits better to other category. Thank you!

Status: Work In Progress

Funding Acknowledgment: So far we are self-funded. Investment round planned for Q1/2018.