Prediction of Individual Speech Outcomes with Cochlear Implants: Limitations and Opportunities for Clinical Translation (2026)
M Campi, A Huber, T Goehring
Under ReviewStatistics · Signal processing · ML
I am Marta!
Doctor in Statistical Signal Processing & Machine Learning
High level experience in the development of statistical and machine learning models for better decision-making processes.
18Research works
18Talks & seminars
14Collaborators
01 / Research
Selected papers
18 research works
A Functional Hierarchy of Semicircular Canal and Otolith Contributions to Early Motor Development (2026)
M Campi, S Wiener, A Maudoux, H Thai-Van
Under ReviewWhich Phonetic Contrasts Recover Information Lost to Aggregate Speech Scoring in Auditory Nerve Disorders: A Computational Framework (2026)
M Campi, E Partouche, G Gerenton, P Avan, C Gaultier
Under ReviewStandardised Hearing Loss Risk Profiles with State-Space Models (2025)
M Campi, GW Peters, P Morvan, M Buhl, H Thai-Van
PublishedBeyond the Audiogram: Using PROMs and AI Modelling to Characterise Age- and Sex-Specific Hearing Loss Needs in a Nationwide Study (2025)
P Morvan, M Campi, GW Peters, H Thai-Van
Under ReviewFrom Hearing Patterns to Functional Outcomes: Quantifying Audiometric Profiles for Precision Hearing Care (2025)
P Morvan, M Campi et al.
Under ReviewDiscrimination loss vs. SRT: A model-based approach towards harmonizing speech test interpretations (2025)
M Buhl, E Kludt, L Schell-Majoor, M Campi, P Avan
Under ReviewAnatomy of Municipal Green Bond Yield Spreads (2025)
MH Sehatpour, M Campi, CS Nikitopoulos, GW Peters, KA Richards
Under ReviewQuantifying The Discriminative Value of Audiological Measurements for Hearing Loss Severity Classification (2025)
M Campi, GW Peters, P Morvan, M Buhl, H Thai-Van
Under ReviewCross-Curve Interest Rate Stress Testing With Endogenous Curve Dynamics (2025)
I Katz, GW Peters, M Campi
Under ReviewAttribute associations of municipal green bond yield spreads: A demand perspective (2025)
MH Sehatpour, M Campi, CS Nikitopoulos, GW Peters, KA Richards
Under ReviewSignature Isolation Forest (2024)
M Campi, G Staerman, GW Peters, T Matsui
arXiv preprint arXiv:2403.04405
AISTAT 2025
PublishedVestibular Impairment and Postural Development in Children With Bilateral Profound Hearing Loss (2024)
SR Wiener-Vacher, M Campi, S Caldani, H Thai-Van
JAMA Network Open 7 (5), e2412846-e2412846
PublishedShades of green: Unveiling the impact of municipal green bonds on the environment (2024)
M Campi, GW Peters, KA Richards
Franklin Open, 100113
PublishedAtaxic speech disorders and Parkinson’s disease diagnostics via stochastic embedding of empirical mode decomposition (2023)
M Campi, G Peters, D Toczydlowska
PLOS One 18 (4), e0284667
PublishedCervical vestibular evoked myogenic potentials in healthy children: Normative values for bone and air conduction (2023)
S Wiener-Vacherm, M Campi, P Boizeau, H Thai-Van
Frontiers in Neurology, Section Neuro-Otology 14
PublishedA Statistical Perspective of the Empirical Mode Decomposition (2022)
M Campi
UCL (University College London)
PhD ThesisMachine Learning Mitigants for Speech Based Cyber Risk (2021)
M Campi, G Peters, N Azzaoui, T Matsui
IEEE Access 9, 136831 - 136860
Published02 / Talks
Presenting my research
18 talks & seminars
Does it Matter Which Fibre You Lose?
12th EAA Annual European Conference on Acoustics and Noise Control Engineering, GrazPrediction of Speech Outcomes with Cochlear Implant Limitations and Opportunities for Clinical Practice
12th International Symposium on Objective Measures in Auditory Implants (OMAI), ZurichBig Data Approaches to Personalised Hearing Care
Seminar at Carl von Ossietzky-Universität OldenburgDetails
Seminar at he Medical Physics Section, Carl von Ossietzky University of Oldenburg in Germany.
Learn morePrediction of Speech Outcomes with Cochlear Implant Limitations and Opportunities for Clinical Practice
Virtual Conference on Computational AudiologyBig Data Approaches to Personalised Hearing Care
6th French-Brazilian Symposium, Department of Speech-Language Pathology and Audiology, Universidade Federal de Minas Gerais (UFMG)Computational Approaches to Hearing Diagnostics Cochlear Implants and Auditory Neuropathy
Computational Linguistics Group, University of ZurichPrediction of Speech Outcomes with Cochlear Implants Limitations and Opportunities for Clinical Practice
Blunch Series, MRC Cognition & Brain Sciences Unit, Cambdrige UniversityTensor Graph Regularized Bootstrap for High-Dimensional Bond Yield Reconstructions
Climate Finance & Risk 2025, Macquarie University City CampusDetails
Invited Keynote Speaker, together with Prof. Gareth W. Peters, at the Climate & Finance Risk 2025 Conference.
Learn moreShades of Green: Unveiling the Impact of Municipal Green Bonds on the Environment
Centre for Climate Risk and Resilience, University Technology SydneyDetails
Invited speaker for the Seminar of the Centre for Climate Risk and Resilience, UTS Business School, University Technology Sydney.
Learn moreAtaxic Speech Disorders and Parkinson’s Disease Diagnostic via Stochastic EMD
University California Santa BarbaraDetails
Invited speaker for the seminar at the Department of Statistics and Applied Probability, University California Santa Barbara.
Learn moreBig Data in Audiology
WCA, World Congress of Audiology 2024Methods for Big Data in Audiology
EFAS, internal workshop for the Congress of European Federation of Audiology Societies 2024From monogenic age-related hearing loss to a battery of objective audiovestibular diagnostic tests
EFAS, 16th Congress of European Federation of Audiology Societies 2023Details
Poster presentation.
Statistical Perspective of the Empirical Mode Decomposition
Télécom Paris, Institut Polytechnique de ParisGreen Bond Risk Performance and Risk Indicators
Heriot-Watt Business SchoolDetails
International Workshop on Green Finance and UK Decarbonization Processes at the Business School of Heriot-Watt University.
Empirical Mode Decomposition for Speaker Verification Applications
Institut of Statistical MathematicsDetails
International Workshop on spatio-temporal statistics methods at ISM in Tokyo.
Properties of Empirical Mode Decomposition
CMStatistics 2017Details
The 10th International Conference of the ERCIM WG on Computational and Methodological Statistics.
Learn moreForecasting Volatility in Financial Markets with High Frequency Data
University of Essex03 / Teaching
From experience to education
Experience · UCL, 2016–2021
University College London (2016-2021)
Teaching Assistant in the Statistical Science and Computer Science departments, instructing courses in Probability, Statistics, and Programming Methods. I received recognition as the Best Tutor within the Computer Science Department (2019) and was awarded the UCL Faculty Education Award. I was also shortlisted for the prestigious UCL Provost's Education Award, competing against more experienced faculty members.
Student supervision
Student Supervision
I supervise/ed graduate and doctoral students across multiple institutions:
PhD students at University of California Santa Barbara and University of Technology Sydney focusing on cross-curve interest rate stress testing and green bond yield analysis.
Master's students at the Hearing Institute from various French Schools such as Sorbonne University (Machine Learning), ECE Paris (Acoustics), and AIX-Marseille Université in different audiological applications.
Teaching philosophy
Philosophy
My approach to teaching is grounded in the belief that theoretical concepts must be connected to real-world applications. I emphasize the development of both technical proficiency and critical thinking skills, encouraging students to move beyond formula memorization to deeper understanding of statistical principles.
I value creating an inclusive learning environment where students from diverse backgrounds can thrive. Drawing from my experience as a professional athlete, I incorporate collaborative problem-solving and team-based learning that prepares students for careers in both academic and industry settings.
I believe in adaptive teaching methods that respond to individual learning styles, using a combination of traditional instruction, interactive demonstrations, and hands-on project work to engage students with different strengths and interests.
04 / Collaborators
Better, together
I maintain active research collaborations across disciplines and institutions worldwide. Along my academic journey, I have had the pleasure of working on several interesting problems with talented researchers.
Gareth W. PetersDuncan Endowed Chair Actuarial & Professor of Statistics for Risk & Insurance.
Hung Thai-VanProfessor of physiology, University of Lyon 1.
Tomoko MatsuiInstitute of Statistical Mathematics.
Tobias GoehringProf. University of Zurich. Lead of Deep Hearing Lab.
Dorota ToczydłowskaSoftware Engineer, NVIDIA, Warsaw.
Nourddine AzzaouiLMBP - Laboratoire de Mathématiques Blaise Pascal.
Mareike BuhlPostDoc, (CERIAH), Institut de l'Audition, Institut Pasteur, Paris.
Ioannis ChalkiadakisFrench National Center for Scientific Research (CNRS).
Guillaume StaermanPostdoc at INRIA (Parietal).
Mohammad Hadi SehatpourUniversity of Technology Sydney (UTS).
Perrine MorvanInstitut Pasteur, CERIAH.
Celine QuinsacProject Manager, Institut Pasteur, CERIAH.
Christina Sklibosios NikitopoulosAssociate Professor, UTS Business School.
Kylie-Anne RichardsSenior Lecturer, UTS Sydney. Head of Investment Research, Director, Future Fund.
05 / About me
A closer look
at my work
UNIVERSITY OF ZURICHINSTITUT PASTEUR · PARIS
UCL | UNIVERSITY COLLEGE LONDON
UoE | UNIVERSITY OF ESSEX
UniGe | UNIVERSITY OF GENOA
Statistics, signal processing and machine learning.
Biography
I am a Marie Skłodowska-Curie Postdoctoral Fellow at the Deep Hearing Lab, University of Zurich and University Hospital Zurich, with a second affiliation at CERIAH, IHU reConnect (Institut Pasteur), Paris. My work sits at the interface of statistics, signal processing, and machine learning, and centres on a single question: what makes hearing outcomes individual, and how far they can be predicted. Within my current project REHEAR (Marie Skłodowska-Curie Actions grant no. 101275781), I develop methods that connect the underlying mechanisms of hearing to the outcomes people experience in practice, with a particular focus on speech understanding in cochlear-implant users and on auditory neuropathy. I work extensively with large-scale audiological data, drawing on statistical signal processing, state-space and time-series modelling, and tools from optimal transport and machine learning to build models that are both accurate and interpretable. These methods are chosen for a reason. Clinical hearing measures are typically aggregated into single scores, discarding the structure that distinguishes one listener from another, and I work on measures and models that preserve it under the constraints of real diagnostic data: few trials, imperfect labels, heterogeneous cohorts. The same methodology transfers beyond hearing, and I have applied it in medical diagnostics, speaker verification and cybersecurity, and the analysis of financial and sustainability data. This is a consistent thread in my work: turning advanced statistical methodology into solutions for complex, structured, real-world data. Trained in Statistical Science and Signal Processing at University College London, I bring a multidisciplinary approach and a track record of international collaboration. Drawing on my background as a professional athlete, I also bring a strategic, results-driven mindset, with an emphasis on adaptability and problem-solving.
Research interests
- Optimal transport (Wasserstein and Gromov-Wasserstein methods)
- Statistical signal processing and non-stationary time-series analy
- State-space modelling
- Copula-based dependence modelling
- Kernel methods (Gaussian processes, SVMs)
- Deep learning models as models of perception (neural-network speech recognition constrained by auditory representations)
- Dimensionality reduction and correlation analysis (kPCA, CCA)
- Signature methods
- Anomaly detection
Application domains
- Cochlear-implant outcome prediction and optimisation
- Auditory neuropathy and hearing-aid optimisation
- Speech recognition and processing
- Medical diagnostics (hearing-loss assessment, Parkinson's disease detection)
- Speaker verification and biometric security
- Financial and sustainability data (mathematical finance, green finance, ESG assessment)