Innovation for Neonatal Care at Philips Research

1. Innovation for Neonatal Care at Philips Research Dr Louis Atallah Senior Scientist, Patient Care Solutions Philips Research, Eindhoven January 2014

2. Philips Research: Innovation around the Globe More than 1500 scientists, cooperating with ~250 universities/institutes Briarcliff Clinical sites Healthcare Lighting Cambridge Home & Oral Healthcare Paris Healthcare Eindhoven Healthcare Lighting Lifestyle Hamburg Healthcare Bangalore Emerging Markets Healthcare Lighting Shanghai Emerging Markets Healthcare Lighting, Lifestyle

3. Patient Care Solutions Enhancing patient care through smart sensor solutions Vision Smart patient care solutions will increasingly support hospital staff in clinical decision making in various care settings. New solutions for lower acuity care will reduce the workload in the ICU. This will also enable use in the home, aiming to improve quality of life and further drive down the cost of healthcare. Mission Through collaboration with clinical experts we deliver innovative technology solutions for Patient Care in the fields of Physiological Monitoring, Respiratory & Emergency care, assisting caregivers both in hospital and home.

4. Why neonatal care? 15 million babies are born preterm every year (WHO) Many babies (not only preterm) have problems and require care in the NICU (Sweden: 10%). Term babies could also be challenging as their problems require quick diagnosis and treatment. Very ill babies need more observation and care and often have longer stays in NICUs than preterm babies.

5. 4 million newborns die every year, most from easily preventable causes Source: World Health Organization 41 41 Source: Lawn, Joy E., Simon Cousens and Jelka Zupan, ‘4 million neonatal deaths; When? Where? Why?', The Lancet, vol. 365, no. 9462, 5 March 2005, p. 895. Neonatal deaths (0–28 days) per 1,000 live births, 2004 30 Mn cases In emerging markets, over 25% of neonates born suffer from hyperbilirubinea

6. NICU Requirements Thermoregulation Humidity Affective Support Calm Environment Lighting Conditions Integration Multidisciplinary team training Baby mannequins Patient Record Medication Control Treatment logging Training Environment Administration Heart Rate Breathing Rate Oxygen Saturation IBP and NIBP CFM Temperature Capnography Blood Gas Body Weight Growth Support Thermoregulation Nutrition Affective Support Calm Environment Monitoring Child Life Support Ventilation/CPAP Oxygen therapy Bio-isolation Access Kangaroo Care Family Diagnosis/Screening Procedures Treatment X-Ray Ultrasound Cranial Ultrasound EEG Physical Examination ROP-screening Intravenous Line Placement Umbilical Artery/Vein Catherization Blood Sampling Diaper Changing Routine Care Endotracheal Intubation Tactile stimulation Resuscitation Medication Phototherapy Brain Cooling Surfactant Inhaled Nitric Oxide * Not exhaustive

7. This presentation Enhancing neonatal care through smart sensor solutions Focus on: Thermoregulation and unobtrusive temperature measurement Unobtrusive ECG monitoring Can monitoring help us predict complications and prevent them earlier?

8. Thermoregulation and core temperature measurement

9. Thermoregulation • Thermoregulation is the ability to balance between heat production and heat loss in order to maintain body temperature within a certain normal range. • Need for thermoregulation: – To protect enzyme function – To maximize metabolic efficiency – To reduce oxygen use – To reduce calorie expenditure – To promote growth and development • New born babies are highly susceptible for heat loss and its adverse consequences (hypothermia) immediately after birth and during first few days of life.

10. Modes of Heat Loss (25 W/m²) (60-80 W/m²) (50 W/m²) (Negligible) Reasons for increased susceptibility to heat Loss in neonates Heat Loss at Birth • Inadequate storage • Poor vasomotor response • High surface area to volume ratio • A lack of thermal insulation • Thin non-keratinized skin How best to measure neonatal temperature?

11. Several methods for measuring temperature Continuous, accurate? Spot-check temperature measurements (axillary) Sensors in the diaper (isolated) Rectal temperatures? Do they really represent core temperature? Esophageal Temperature: • Surrogate for core temperature • Temperature changes can be sudden and dramatic, not reflected by spot-check measurements. Philips InnerSense Esophageal Temperature Probe+ Feeding Tube

12. Philips InnerSense Esophageal Temperature Probe+ Feeding Tube Unobtrusive core temperature measurement A feeding tube containing temperature sensors to reliably measure core temperature. Temperature can be measured, displayed on the patient monitor and used for trend analysis. Recent study we did on 12 babies showed that this sensor is more stable than a skin sensor placed in the diaper especially during motion and kangaroo care.

13. Contactless ECG

14. NICU| Environment • Many procedures are necessary and standard, and have reduced viability to 24 weeks. However, these same procedures can have side effects! • Skin is incompletely developed (not until 34 weeks that the skin is functionally competent). Scars prevail in 10% of neonates in the NICU. • ECG is important in observing neonatal conditions and can indicate life threatening events such as sepsis (infection). • A closer look…. Adhesive electrodes

15. Embedded neonatal monitoring| ENEMO • Neonate: improved comfort (less scars and pain) • Caregiver: less preparation time • Parents: improved parent-child bonding

16. The trial| Contactless capacitive ECG • Trial at the Maxima Medical Centre, Veldhoven, NL. Clinical data of 20 neonates (100 hours of data) + Reference- Camera- Notes. • Primary objective: Evaluate the performance of capacitive ECG against adhesive ECG. • Secondary objective: Learn about conditions in the NICU. What are the (un)desired conditions for an optimal recording?

17. Motion levels|babies are different 18% 37% 13% 56% 69% 7% Total time 3.75 hours Total time: 5.7 hours

18. Framework| The Vectorcardiogram • A graphic representation of the magnitude and direction of the electrical currents in the heart as a vector loop. Vectorcardiogram (VCG)

19. Framework| ECG reconstruction Reference ECG Capacitive ECG Project on standard Einthoven leads Pre-processing/ subtracting motion artefacts VCG construction VCG Projection ECG R-peak identification using wavelet techniques Kalman filtering of channels (knowing beat locations) VCG Projection (again)

20. Framework| In real time…

21. Results| Instantaneous Heart Rate coverage Coverage Low number of layers, mostly chest position, good sensor alignment Coverage: Total time with a good match. Peak detection used to calculate true positives and negatives leading to Sensitivity and PPV. Reference ECG Capacitive ECG Percent coverage Percent coverage Very bad sensor alignment, side positioning Baby number Baby number

22. Results| Effect of layers and positioning Baby 9 Bad conditions Good conditions 4 layers of cloth, side positioning 1 layer removed, chest positioning

23. Results| When it’s good it’s really good • Baby 1: Overall Coverage: 86%, max time to wait for a good iHR: 4.1 sec. Percentage No Motion Low Motion Intervention

24. Results| ECG shape analysis • Baby 1 R Reference P T Q S R Capacitive T P S For more results see: PR-TN 2012/00614 Capacitive ECG sensing in the NICU - design and results of the clinical feasibility study Q

25. NICU| ECG relevance • 7 Dutch and 10 international NICUs (UK, Spain, Ireland, USA, Belgium) responded to a survey we sent. • ECG an important tool for screening: 90% of participants check it regularly whereas 10% of participants check it only once in a while. • In order of importance: HR, HR variation, Respiration rate, QRS shape variation. • Instantaneous heart rate (iHR) an important parameter for Sepsis detection1. Sepsis is the most common cause of death for infants beyond 1 week of age. 1. Early non invasive diagnosis of sepsis in preterm infants (Griffin Ped Research 2003; Griffin Pediatrics 2005; Moorman IEEE 2006, Fairchild 2012).

26. Other players| Benchmarking Novelty Aspects Adaptive framework that can deal with motion and artefacts Unobtrusive embedding with no direct contact Neonatal monitoring Strong IP position 1, 2 3 4 • Eilebrecht et al. Cardiovascular Engineering and Technology 2012. • Eilebreacht et al. AutomobiltechnischeZeitschrift ATZ, 2011. • Rasenack et al. Clinical Research in Cardiology 2012. • Oehler et al. Phys. Measurement 2008.

27. What can we do with instantaneous heart rate?

28. Detection of neonatal sepsis using heart rate characteristics Background Late-onset sepsis (onset > 72 hr after birth): 30% of preterm very low birth weight infants experience sepsis during their NICU stay (Stoll 2002) Sepsis is the most common cause of death among preterm infants beyond the first week of life. Heart rate characteristics monitoring (e.g changes in heart rate pattern) have shown to aid in the early non invasive diagnosis of sepsis in preterm infants (Griffin Ped Research 2003; Griffin Pediatrics 2005; Moorman IEEE 2006).

29. Detection of neonatal sepsis using heart rate characteristics • Background Healthy infant loss of variability & increase of transient decelerations, not exceeding alarm limits! Septic infant Griffin Pediatrics 2001 KovatchevPed Research 2003

30. The HRCiis derivedfromregressionmodeling and usesmeasures of standarddeviation(SD), sample symmetry(R1 and R2), and SampEnto estimate the risk of upcoming sepsis and sepsis-likeillness. • The formulafor the HRC index is: HRCi= [exp(A) / 1+exp(A)] where: A = intercept + β1(SD) + β2(R1) + β3(R2) + β4(SampEn) • The intercept and coefficients β are estimatedfromepidemiologic data on sepsis and sepsis-likeillness in the NICU (training set) Detection of neonatal sepsis using heart rate characteristics • Algorithm

31. Detection of neonatal sepsis using heart rate characteristics • HRCi as relative risk HRCipercentilescorrelatewith risk for sepsis. Patients are classifiedinto 3 risk categories. Low risk is defined as HRCi < 75th percentile (score <1). Medium risk (lightgrayboxes) includesinfantswithHRCi in the 75th to 90th percentile (score 1–2). The highrisk HRC group (darkgray box) has HRCigreaterthan 90th percentile (score >3). Reference: Griffin Pediatrics 2005

32. Detection of neonatal sepsis using heart rate characteristics • Example Increase of HRCi 24 hrbeforeblood culture & start of antibiotics

33. In conclusion…

34. Conclusion Neonates are a very vulnerable population. Research in this area could benefit a lot from: • Unobtrusive and pervasive sensor systems. • Support for neonatal comfort, sleep and growth • Intelligent solutions using the data acquired for early prediction of deterioration. • Support for nurses and clinicians for more efficient and easier care.

35. Fulfilling our ambitions in Mother and Child CareHolisticcomprehensive portfolio Avent Home apnea monitoring OB Ultrasound Neonatal Respiratory Support Fetal and maternal monitoring Neonatal patient monitoring Jaundice diagnostic and therapy Developmental Care Positioning aids Feeding and Specialty products Incubators and warmers 35

36. People centered design How we work Research Onsite observation, shadowing, stakeholder interviews Analysis Understand workflows & issues Uncover insights & opportunities Concept creation New materials, manufacturing & marketing feasibility Validation with clinical experts, clinical trials with end users

37. Acknowledgments | The team • Team • Mohammed Meftah – Philips Research • MartijnSchellekens- Philips Research • Aline Serteyn- TU/e • Rik Vullings-TU/e • Jan Bergmans –TU/e • Sidarto Bambang Oetomo- MMC/TU/e • Astrid Osagiator-MMC • Edwin Bongers-Philips Research • Eefje Arts-Hornix- Philips Research • Marjolein van Lieshout-Philips Research • Vanessa Sattele-Philips Design • Anton Janssen-Philips Research Philips Business • Siegfried Kaestle • Amy Phillips • Leslie Altimier Collaborators • Wei Chen- TU/e • LoeFeijs-TU/e • MishaCroes-TU/e • DominikaPotuzakova-TU/e • Joanna Pilarczyk-TU/e • Chris Nederhorst-RHF • Hugo Zijlmans- AME • Jos Bax (PINS) • Mark Bogers (PINS)