top of page

Project goal

Design and implement a new method for early detection of neonatal sepsis for home based settings in developing countries.

 

 

 

 

 

Background

Problem Description:

  • Neonatal sepsis is the whole-body inflammatory response to infections.

  • The rapid progression of sepsis can occur in a matter of hours, making early detection of neonatal sepsis vital in ensuring neonates receive medical treatment.

(read more)

 

 

 

 

 

 

 

 

 

 

 

 

Fig. 1. Example of physiological changes that occur in a septic neonate. Many of these are either late-stage symptoms or difficult to notice in practice.

 

Need statement

Community based healthcare workers need a method to detect early signs of neonatal sepsis following a delivery in low resource settings to help neonates’ family members and healthcare workers seek timely interventions, thus reducing sepsis-related neonatal mortality.

(read more)

 

The Team

The team is composed of undergraduate Biomedical Engineering students and professionals from different divisions at Johns Hopkins University.

 

Our advisors are faculty across the Johns Hopkins CBID program, school of public health, the Johns Hopkins Hospital, and people currently working in developing countries.

(read more)

Solution Concept and Prototype

The solution includes three components. The SepSock contains the heart rate sensor. The SepScore algorithm provides a score that expresses whether the neonate developed sepsis. The third component is the processing device that can collect the data, run the algorithm and be connected to a basic phone in order to send the SepScore.

 

 

 

 

 

 

 

Fig 2. Overview of the device. Shown are the SepBox component and the heart rate sensor.


Physical Device and Algorithm

  • The prototype is a small plastic box incorporating a microprocessor and heart rate sensor

 

 

 

 

 

 

 

 

 

 

Fig. 3. SepBox components labeled. The orange is a hard plastic casing; the 2.5 mm audio jack is compatible with basic bar phones.

 

  • The algorithm is a generalized linear model that searches for subtle sepsis-indicating changes in heart rate characteristics

 

 

 

 

 

 

 

Fig. 4. Illustration of the algorithm. Periodically, the measured heart rate will be quantified by a number of features, which are used in discriminating sepsis vs. non-sepsis.

(read more)

 

Testing

  • To verify the accuracy of our sensor, we tested it during rest, exercise, and various movement conditions, and its results were compared to the data collected from a gold standard (a pulse oximeter).

  • This simulates the kicking motion of a neonate and various levels of cardiovascular stimulation.

 

 

 

 

 

 

 

 

 

 

 

Fig 5. Heart rate measurements from both sensors during rest (left) and 90 move/min hand waving (right)

(read more)

Solution landscape

  • There are no methods specific to detecting neonatal sepsis in low resource settings.

  • Current gold standard is to culture blood for pathogenic organisms, but this can take up to 48 hours, which is an unacceptably long delay in treatment.

  • The following table compares NeoSED’s device to its equivalent devices in the United States.

 

 

 

 

 

 

 

 

 

 

 

Fig 6. Comparison of NeoSED to other scoring systems and wearable devices. In short, although the solution landscape includes similar technologies, none of the prior art is suitable for the specific global health target of our system.

 

Implementation pathway

Next steps: Collection of neonatal data at Johns Hopkins Hospital

Adoption pathway: Preliminary small scale dissemination in Bangladesh, followed by larger scale deployment in India

Regulatory classification: FDA approval for a Class II device with predicate

IP protection plan: design patent for SepSock, utility patent for the SepScore system, copyright for the manual and a trademark

(read more)

 

News

Milestones

  • First successful iteration of SepScore (December 6th, 2015)

  • First incorporation of phone communication in physical device. (February 20th, 2016)

 

 

 

 

 

 

 

 

 

 

Fig 7. An early prototype. This prototype had messaging capabilities, but the algorithm had to be run in an external computer.

 

Obstacles

  • Lack of neonatal data for the development of the algorithm; an IRB is in preparation to collect neonatal data from the Johns Hopkins Hospital

 

Awards

  • NeoSED presented as 1 of the 10 finalists in the Bay Area Global Health Innovation Challenge in Berkeley. (April 8-9, 2016)

  • NeoSED was a finalist at the Johns Hopkins Business Plan Competition in the Global Health Category. (April 1st, 2016)

(read more)
 

 

bottom of page