Machine Learning From Fetal Flow Waveforms to Predict Adverse Perinatal Outcomes

NCT ID: NCT03398551

Last Updated: 2018-01-12

Basic Information

• Recruitment Status: UNKNOWN
• Total Enrollment: 525 participants
• Study Classification: OBSERVATIONAL
• Study Start Date: 2018-02-28
• Study Completion Date: 2018-12-31

Brief Summary

The aim of this study is to get a proof of concept for using a computational model of fetal haemodynamics, combined with machine learning based on Doppler patterns of the fetal cardiovascular, cerebral and placental flows, to identify those at increased risk of adverse perinatal outcomes such as stillbirth, perinatal mortality and other neonatal morbidities.

We will also compare the sensitivity and specificity of UmbiFlow device with the machine learning model in predicting adverse perinatal outcomes

Detailed Description

Pakistan is one of the countries where stillbirth rate (43/1000 total births) and neonatal mortality rate (55/1000 live births) are among the highest in the world. The figures for perinatal mortality remained unchanged between Pakistan Demographic and Health Survey (PDHS) 1990-91 and PDHS 2012-13. It is hence important to evaluate new techniques to identify high-risk pregnant women who can then subsequently receive personalized care to improve perinatal outcomes. Fetal blood flow waveforms assessed via Doppler ultrasound to indicate wellbeing of the fetus have been in practice for nearly four decades. Abnormal blood flow patterns in the fetal circulation detected by Doppler ultrasound may indicate poor fetal prognosis. These tracings will then be incorporated in a computational model and combined with machine learning, will be used to predict adverse perinatal outcomes.

This will be a prospective one group cohort study for a period of twelve months which will be based in Ibrahim Hyderi, a peri-urban settlement of approximately 70,000 on the south east of Karachi. The sample size for this study has been calculated based on the outcomes of the pregnancy and was estimated to be 525 pregnant women.

Recruitment - Household surveillance will be done by community health workers (CHWs) in the catchment area in a sequential manner to identify pregnant women. Eligible women between 22-34 weeks of gestation on last menstrual period (LMP) will then be explained regarding the purpose of the study and invited for an ultrasound scan.

First and enrollment visit - The first visit window for the ultrasound scan will be between 22-26 weeks of gestation. This window is selected because pregnancies at risk may be better identified and also visualization of fetal cardiac flow is possible. Based on the rigorous surveillance, we anticipate that most women will be in this window. The following information will be collected:

• Socio-demographic information
• Medical history, including exposure to tobacco, gravidity, parity, past obstetric history (including number of miscarriages, stillbirths or neonatal deaths) and comorbid conditions (pre-eclampsia, diabetes mellitus)
• Maternal anthropometry (height, weight and mid-upper arm circumference)
• Maternal blood pressure
• Hemoglobin of the mother
• Symphysiofundal height will be measured
• Standard obstetric Doppler evaluation including the four primary ultrasound measures of fetal growth-head circumference, biparietal diameter, abdominal circumference, and femur length is routinely performed for all pregnant women presenting at the center. These parameters will be used to estimate fetal weight (automatically calculated by the machine).
• Focused Fetal Doppler Echocardiography - These would include transabdominal measurements of the Doppler velocity of the umbilical artery, the middle cerebral artery (MCA), ductus arteriosus, fetal aortic isthmus (AoI) and left and right ventricular outflow tracts
• UmbiFlow - Umbilical artery continuous wave Doppler image will also be obtained on these women. The device will generate graphs for the flow pattern and calculate the resistance index for the flow.

Second visit - The second visit window of ultrasound scan will be between 30-34 weeks of gestation. This will be applicable for all women who have had their first scan during 22-26 weeks of gestation as well as newly enrolled women during 30-34 weeks of gestational age. Cardiotocography (CTG) will be performed for all recruited women during the second visit followed by the Doppler examination. The same procedures as described in the first visit will be done.

Follow up - We will obtain contact information (phone number) from all enrolled women. A trained research midwife will be responsible for making phone calls, as well as performing household visits at pre-specified intervals for immediate and early identification of the pregnancy outcome. A team of CHWs will perform home visits to the baby at post-natal day 1 and 6 and questions related to delayed cry, neonatal sepsis and hospital course (if any) will be collected and the weight and length of the baby will be recorded.

Verbal autopsy - The VA team will visit the family after the grieving period (at least 14 days after the death). VA will be performed using a structured tool (http://www.who.int/healthinfo/statistics/WHO_VA_2012_RC1_Instrument.pdf ).

Conditions

Perinatal Mortality; Neonatal Morbidities

Study Design

• Observational Model Type: COHORT
• Study Time Perspective: PROSPECTIVE

Eligibility Criteria

Inclusion Criteria

• Pregnant woman coming to the ultrasound clinic between 22-34 weeks of gestation.
• Written informed consent
• Resident of the study area

Exclusion Criteria

• Multiple gestation
• Known congenital anomaly in the fetus or newborn
• Refusal for the ultrasound
• Poor echocardiographic images for Doppler acquisition

• Eligible Sex: FEMALE
• Accepts Healthy Volunteers: Yes

Sponsors

Universitat Pompeu Fabra

OTHER

Sponsor Role: collaborator

Aga Khan University

OTHER

Sponsor Role: lead

Responsible Party

Zahra Hoodbhoy

Dr Zahra Hoodbhoy

Responsibility Role: PRINCIPAL_INVESTIGATOR

Central Contacts

Zahra Hoodbhoy

Role: CONTACT

zahra.hoodbhoy@aku.edu

92-21-34930051 ext. 4126

Babar Hasan

Role: CONTACT

babar.hasan@aku.edu

92-21-34930051 ext. 4364

Other Identifiers

4872-Ped-ERC-17

Identifier Type: -

Identifier Source: org_study_id