Predicting neurodevelopmental outcomes from neonatal cortical microstructure: A conceptual replication study

Andrea Gondová; Sara Neumane; Yann Leprince; Jean François Mangin; Tomoki Arichi; Jessica Dubois

doi:10.1016/j.ynirp.2023.100170

Predicting neurodevelopmental outcomes from neonatal cortical microstructure: A conceptual replication study

Andrea Gondová^*, Sara Neumane, Yann Leprince, Jean François Mangin, Tomoki Arichi, Jessica Dubois

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

Abstract

Machine learning combined with large-scale neuroimaging databases has been proposed as a promising tool for improving our understanding of the behavioural emergence and early prediction of the neurodevelopmental outcome. A recent example of this strategy is a study by Ouyang et al. (2020) which suggested that cortical microstructure quantified by diffusion MRI through fractional anisotropy (FA) metric in preterm and full-term neonates can lead to effective prediction of language and cognitive outcomes at 2 years of corrected age as assessed by Bayley Scales of Infant and Toddler Development, Third Edition (BSID-III) composite scores. Given the important need for robust and generalisable tools which can reliably predict the neurodevelopmental outcome of preterm infants, we aimed to replicate the conclusions of this work using a larger independent dataset from the developing Human Connectome Project dataset (dHCP, third release) with early MRI and BSID-III evaluation at 18 months of corrected age. We then aimed to extend the validation of the proposed predictive pipeline through the study of different cohorts (the largest one included 295 neonates, with gestational age between 29 and 42 week and post-menstrual age at MRI between 31 and 45 weeks). This allowed us to evaluate whether some limitations of the original study (mainly small sample size and limited variability in the input and output features used in the predictive models) would influence the prediction results. In contrast to the original study that inspired the current work, our prediction results did not outcompete the random levels. Furthermore, these negative results persisted even when the study settings were expanded. Our findings suggest that the cortical microstructure close to birth described by DTI-FA measures might not be sufficient for a reliable prediction of BSID-III scores during toddlerhood, at least in the current setting, i.e. generally older cohorts and a different processing pipeline. Our inability to conceptually replicate the results of the original study is in line with the previously reported replicability issues within the machine learning field and demonstrates the challenges in defining the good set of practices for the implementation and validation of reliable predictive tools in the neurodevelopmental (and other) fields.

Original language	English
Article number	100170
Journal	Neuroimage: Reports
Volume	3
Issue number	2
Early online date	14 Apr 2023
DOIs	https://doi.org/10.1016/j.ynirp.2023.100170
Publication status	Published - Jun 2023

Keywords

Brain development
DTI (Diffusion tensor imaging)
Generalisability
ML (machine learning)
Neonates
Prediction
Prematurity

Access to Document

10.1016/j.ynirp.2023.100170

Cite this

@article{2328eb69568c4da1a10665421ff4b647,

title = "Predicting neurodevelopmental outcomes from neonatal cortical microstructure: A conceptual replication study",

abstract = "Machine learning combined with large-scale neuroimaging databases has been proposed as a promising tool for improving our understanding of the behavioural emergence and early prediction of the neurodevelopmental outcome. A recent example of this strategy is a study by Ouyang et al. (2020) which suggested that cortical microstructure quantified by diffusion MRI through fractional anisotropy (FA) metric in preterm and full-term neonates can lead to effective prediction of language and cognitive outcomes at 2 years of corrected age as assessed by Bayley Scales of Infant and Toddler Development, Third Edition (BSID-III) composite scores. Given the important need for robust and generalisable tools which can reliably predict the neurodevelopmental outcome of preterm infants, we aimed to replicate the conclusions of this work using a larger independent dataset from the developing Human Connectome Project dataset (dHCP, third release) with early MRI and BSID-III evaluation at 18 months of corrected age. We then aimed to extend the validation of the proposed predictive pipeline through the study of different cohorts (the largest one included 295 neonates, with gestational age between 29 and 42 week and post-menstrual age at MRI between 31 and 45 weeks). This allowed us to evaluate whether some limitations of the original study (mainly small sample size and limited variability in the input and output features used in the predictive models) would influence the prediction results. In contrast to the original study that inspired the current work, our prediction results did not outcompete the random levels. Furthermore, these negative results persisted even when the study settings were expanded. Our findings suggest that the cortical microstructure close to birth described by DTI-FA measures might not be sufficient for a reliable prediction of BSID-III scores during toddlerhood, at least in the current setting, i.e. generally older cohorts and a different processing pipeline. Our inability to conceptually replicate the results of the original study is in line with the previously reported replicability issues within the machine learning field and demonstrates the challenges in defining the good set of practices for the implementation and validation of reliable predictive tools in the neurodevelopmental (and other) fields.",

keywords = "Brain development, DTI (Diffusion tensor imaging), Generalisability, ML (machine learning), Neonates, Prediction, Prematurity",

author = "Andrea Gondov{\'a} and Sara Neumane and Yann Leprince and Mangin, {Jean Fran{\c c}ois} and Tomoki Arichi and Jessica Dubois",

note = "Funding Information: JFM and JD are supported by the Fondation Paralysie C{\'e}r{\'e}brale for the ENSEMBLE project. Funding Information: The developing Human Connectome Project was funded by the European Research Council under the European Union Seventh Framework Programme ( FP/2007–2013 )/ERC Grant Agreement no. 319456 . Funding Information: AG is supported by the CEA NUMERICS program. This project has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 800945 — NUMERICS — H2020-MSCA-COFUND-2017 . Funding Information: TA is supported by a MRC Clinician Scientist Fellowship [ MR/P008712/1 ] and MRC Transition Support Award [ MR/V036874/1 ]. Funding Information: JD received support from the Fondation M{\'e}disite (under the aegis of the Fondation de France , grant FdF-18-00092867 ) and the IdEx Universit{\'e} de Paris ( ANR-18-IDEX-0001 ). Funding Information: SN was supported by a postdoctoral fellowship from the Bettencourt Schueller Foundation ( www.fondationbs.org ). Publisher Copyright: {\textcopyright} 2023",

year = "2023",

month = jun,

doi = "10.1016/j.ynirp.2023.100170",

language = "English",

volume = "3",

journal = "Neuroimage: Reports",

issn = "2666-9560",

publisher = "Elsevier",

number = "2",

}

TY - JOUR

T1 - Predicting neurodevelopmental outcomes from neonatal cortical microstructure

T2 - A conceptual replication study

AU - Gondová, Andrea

AU - Neumane, Sara

AU - Leprince, Yann

AU - Mangin, Jean François

AU - Arichi, Tomoki

AU - Dubois, Jessica

N1 - Funding Information: JFM and JD are supported by the Fondation Paralysie Cérébrale for the ENSEMBLE project. Funding Information: The developing Human Connectome Project was funded by the European Research Council under the European Union Seventh Framework Programme ( FP/2007–2013 )/ERC Grant Agreement no. 319456 . Funding Information: AG is supported by the CEA NUMERICS program. This project has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No 800945 — NUMERICS — H2020-MSCA-COFUND-2017 . Funding Information: TA is supported by a MRC Clinician Scientist Fellowship [ MR/P008712/1 ] and MRC Transition Support Award [ MR/V036874/1 ]. Funding Information: JD received support from the Fondation Médisite (under the aegis of the Fondation de France , grant FdF-18-00092867 ) and the IdEx Université de Paris ( ANR-18-IDEX-0001 ). Funding Information: SN was supported by a postdoctoral fellowship from the Bettencourt Schueller Foundation ( www.fondationbs.org ). Publisher Copyright: © 2023

PY - 2023/6

Y1 - 2023/6

N2 - Machine learning combined with large-scale neuroimaging databases has been proposed as a promising tool for improving our understanding of the behavioural emergence and early prediction of the neurodevelopmental outcome. A recent example of this strategy is a study by Ouyang et al. (2020) which suggested that cortical microstructure quantified by diffusion MRI through fractional anisotropy (FA) metric in preterm and full-term neonates can lead to effective prediction of language and cognitive outcomes at 2 years of corrected age as assessed by Bayley Scales of Infant and Toddler Development, Third Edition (BSID-III) composite scores. Given the important need for robust and generalisable tools which can reliably predict the neurodevelopmental outcome of preterm infants, we aimed to replicate the conclusions of this work using a larger independent dataset from the developing Human Connectome Project dataset (dHCP, third release) with early MRI and BSID-III evaluation at 18 months of corrected age. We then aimed to extend the validation of the proposed predictive pipeline through the study of different cohorts (the largest one included 295 neonates, with gestational age between 29 and 42 week and post-menstrual age at MRI between 31 and 45 weeks). This allowed us to evaluate whether some limitations of the original study (mainly small sample size and limited variability in the input and output features used in the predictive models) would influence the prediction results. In contrast to the original study that inspired the current work, our prediction results did not outcompete the random levels. Furthermore, these negative results persisted even when the study settings were expanded. Our findings suggest that the cortical microstructure close to birth described by DTI-FA measures might not be sufficient for a reliable prediction of BSID-III scores during toddlerhood, at least in the current setting, i.e. generally older cohorts and a different processing pipeline. Our inability to conceptually replicate the results of the original study is in line with the previously reported replicability issues within the machine learning field and demonstrates the challenges in defining the good set of practices for the implementation and validation of reliable predictive tools in the neurodevelopmental (and other) fields.

AB - Machine learning combined with large-scale neuroimaging databases has been proposed as a promising tool for improving our understanding of the behavioural emergence and early prediction of the neurodevelopmental outcome. A recent example of this strategy is a study by Ouyang et al. (2020) which suggested that cortical microstructure quantified by diffusion MRI through fractional anisotropy (FA) metric in preterm and full-term neonates can lead to effective prediction of language and cognitive outcomes at 2 years of corrected age as assessed by Bayley Scales of Infant and Toddler Development, Third Edition (BSID-III) composite scores. Given the important need for robust and generalisable tools which can reliably predict the neurodevelopmental outcome of preterm infants, we aimed to replicate the conclusions of this work using a larger independent dataset from the developing Human Connectome Project dataset (dHCP, third release) with early MRI and BSID-III evaluation at 18 months of corrected age. We then aimed to extend the validation of the proposed predictive pipeline through the study of different cohorts (the largest one included 295 neonates, with gestational age between 29 and 42 week and post-menstrual age at MRI between 31 and 45 weeks). This allowed us to evaluate whether some limitations of the original study (mainly small sample size and limited variability in the input and output features used in the predictive models) would influence the prediction results. In contrast to the original study that inspired the current work, our prediction results did not outcompete the random levels. Furthermore, these negative results persisted even when the study settings were expanded. Our findings suggest that the cortical microstructure close to birth described by DTI-FA measures might not be sufficient for a reliable prediction of BSID-III scores during toddlerhood, at least in the current setting, i.e. generally older cohorts and a different processing pipeline. Our inability to conceptually replicate the results of the original study is in line with the previously reported replicability issues within the machine learning field and demonstrates the challenges in defining the good set of practices for the implementation and validation of reliable predictive tools in the neurodevelopmental (and other) fields.

KW - Brain development

KW - DTI (Diffusion tensor imaging)

KW - Generalisability

KW - ML (machine learning)

KW - Neonates

KW - Prediction

KW - Prematurity

UR - http://www.scopus.com/inward/record.url?scp=85153530528&partnerID=8YFLogxK

U2 - 10.1016/j.ynirp.2023.100170

DO - 10.1016/j.ynirp.2023.100170

M3 - Article

AN - SCOPUS:85153530528

SN - 2666-9560

VL - 3

JO - Neuroimage: Reports

JF - Neuroimage: Reports

IS - 2

M1 - 100170

ER -

Predicting neurodevelopmental outcomes from neonatal cortical microstructure: A conceptual replication study

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this