-
Detecting the Severity of Major Depressive Disorder from Speech: A Novel HARD-Training Methodology
Authors:
Edward L. Campbell,
Judith Dineley,
Pauline Conde,
Faith Matcham,
Femke Lamers,
Sara Siddi,
Laura Docio-Fernandez,
Carmen Garcia-Mateo,
Nicholas Cummins,
the RADAR-CNS Consortium
Abstract:
Major Depressive Disorder (MDD) is a common worldwide mental health issue with high associated socioeconomic costs. The prediction and automatic detection of MDD can, therefore, make a huge impact on society. Speech, as a non-invasive, easy to collect signal, is a promising marker to aid the diagnosis and assessment of MDD. In this regard, speech samples were collected as part of the Remote Assess…
▽ More
Major Depressive Disorder (MDD) is a common worldwide mental health issue with high associated socioeconomic costs. The prediction and automatic detection of MDD can, therefore, make a huge impact on society. Speech, as a non-invasive, easy to collect signal, is a promising marker to aid the diagnosis and assessment of MDD. In this regard, speech samples were collected as part of the Remote Assessment of Disease and Relapse in Major Depressive Disorder (RADAR-MDD) research programme. RADAR-MDD was an observational cohort study in which speech and other digital biomarkers were collected from a cohort of individuals with a history of MDD in Spain, United Kingdom and the Netherlands. In this paper, the RADAR-MDD speech corpus was taken as an experimental framework to test the efficacy of a Sequence-to-Sequence model with a local attention mechanism in a two-class depression severity classification paradigm. Additionally, a novel training method, HARD-Training, is proposed. It is a methodology based on the selection of more ambiguous samples for the model training, and inspired by the curriculum learning paradigm. HARD-Training was found to consistently improve - with an average increment of 8.6% - the performance of our classifiers for both of two speech elicitation tasks used and each collection site of the RADAR-MDD speech corpus. With this novel methodology, our Sequence-to-Sequence model was able to effectively detect MDD severity regardless of language. Finally, recognising the need for greater awareness of potential algorithmic bias, we conduct an additional analysis of our results separately for each gender.
△ Less
Submitted 25 May, 2023; v1 submitted 2 June, 2022;
originally announced June 2022.
-
Remote smartphone-based speech collection: acceptance and barriers in individuals with major depressive disorder
Authors:
Judith Dineley,
Grace Lavelle,
Daniel Leightley,
Faith Matcham,
Sara Siddi,
Maria Teresa Peñarrubia-María,
Katie M. White,
Alina Ivan,
Carolin Oetzmann,
Sara Simblett,
Erin Dawe-Lane,
Stuart Bruce,
Daniel Stahl,
Yatharth Ranjan,
Zulqarnain Rashid,
Pauline Conde,
Amos A. Folarin,
Josep Maria Haro,
Til Wykes,
Richard J. B. Dobson,
Vaibhav A. Narayan,
Matthew Hotopf,
Björn W. Schuller,
Nicholas Cummins,
The RADAR-CNS Consortium
Abstract:
The ease of in-the-wild speech recording using smartphones has sparked considerable interest in the combined application of speech, remote measurement technology (RMT) and advanced analytics as a research and healthcare tool. For this to be realised, the acceptability of remote speech collection to the user must be established, in addition to feasibility from an analytical perspective. To understa…
▽ More
The ease of in-the-wild speech recording using smartphones has sparked considerable interest in the combined application of speech, remote measurement technology (RMT) and advanced analytics as a research and healthcare tool. For this to be realised, the acceptability of remote speech collection to the user must be established, in addition to feasibility from an analytical perspective. To understand the acceptance, facilitators, and barriers of smartphone-based speech recording, we invited 384 individuals with major depressive disorder (MDD) from the Remote Assessment of Disease and Relapse - Central Nervous System (RADAR-CNS) research programme in Spain and the UK to complete a survey on their experiences recording their speech. In this analysis, we demonstrate that study participants were more comfortable completing a scripted speech task than a free speech task. For both speech tasks, we found depression severity and country to be significant predictors of comfort. Not seeing smartphone notifications of the scheduled speech tasks, low mood and forgetfulness were the most commonly reported obstacles to providing speech recordings.
△ Less
Submitted 30 August, 2021; v1 submitted 17 April, 2021;
originally announced April 2021.
-
The distribution of maser stars in the inner Milky Way: the effect of a weak, rotating bar
Authors:
H. J. Habing,
M. N. Sevenster,
M. Messineo,
G. van de Ven,
K. Kuijken,
:,
the ROTBAR consortium
Abstract:
We derive the distribution of maser stars in the inner Milky Way (MW) based on an analysis of lV-diagrams (lVd) for two samples of maser stars: 771 OH/IR stars and 363 SiO-maser stars. They are all close to the plane of the MW and have long. from -45 to +45deg. The two lVds are compared and found to be very similar. They also compare well with the lVd of interstellar CO, but there are significan…
▽ More
We derive the distribution of maser stars in the inner Milky Way (MW) based on an analysis of lV-diagrams (lVd) for two samples of maser stars: 771 OH/IR stars and 363 SiO-maser stars. They are all close to the plane of the MW and have long. from -45 to +45deg. The two lVds are compared and found to be very similar. They also compare well with the lVd of interstellar CO, but there are significant differences in detail between the stellar lVds and that of the ISM. Based on the qualitative discussion we divide the lVds into seven areas. In each area we compare the number of stars observed with those predicted by an assumed set of orbits in a galactic potential. This potential is axially symmetric but a weak rotating bar has been added. We conclude that the maser stars move on almost circular orbits outside of about 3.5 kpc, but that the orbits become more and more elongated when one goes deep inside our MW. We find a strong effect of the Corotation (CR) resonance (res) at 3.3 kpc, we see a small but noticeable effect of the Outer Lindblad res at 5 kpc and no effect of the Inner Lindblad res at r=0.8 kpc. We find a set of 6 groups of orbits that together predict counts in agreement with the counts of stars observed. We then calculate the trajectory of each orbit and so find the distribution of the maser stars in the plane of the MWG. This distribution has two new (but not unexpected) features. The first is a bar-like distribution within 2 kpc from the GC outlined. These orbits explain the high-vel stars near l=0deg in the forbidden and the permitted quadrants. The second feature are two "croissant"-like voids in the distribution close to the CR radius (3.3 kpc), which are the consequence of the presence of the CR res. We find excellent agreement with an earlier reconstruction by Sevenster (1999).
△ Less
Submitted 24 August, 2006;
originally announced August 2006.
