Vocal Pattern Assessment as a New Key to Identifying Rare Syndromes
- Conditions
- Rare DiseasesCongenital Disorders
- Interventions
- Other: voice recordings
- Registration Number
- NCT05955794
- Brief Summary
Primary Study Objective The primary objective of the study is the definition of distinct vocal phenotypes and the development of an Explained Decision Support System (DSS) for the automatic detection of vocal patterns in relation to the syndrome from which the patients suffer
Secondary:
1. Perceptual and acoustic analysis of voice recordings
2. Development of a voice recording collection system.
- Detailed Description
Genetic syndromes have been extensively studied and numerous research studies have enabled a better definition of their clinical manifestations, natural history and aetiopathogenetic mechanisms. In these multisystem conditions, some relevant but as yet unexplored aspects need to be clarified, and one of these aspects is the characterisation of vocal production. To date, more than 240 genetic syndromes have distinctive voice quality abnormalities that are significant enough to be considered diagnostic indicators. However, with the exception of Down syndromes, X-fragile, Williams and velocardiofacial syndromes, no scientific studies focusing on vocal analysis have been conducted on patients with genetic syndromes in order to obtain a complete and objective characterisation of vocal characteristics.
Due to the sophisticated human auditory apparatus, perceptual analysis is the basic approach for assessing voice quality. However, an objective assessment of voice quality is crucial in order to minimise errors due to perceptual and, consequently, individual analysis. Currently, the most widely used tool by researchers is Praat, although its use, being not very intuitive, can be challenging for those with little computer experience. Consequently, the use of the default settings (adult male voice) may give somewhat misleading results. To partially overcome some of the above-mentioned problems, a new user-friendly software called Biovoice has recently been developed. Recently, a number of new measurement methods have been designed to assess vocal characteristics, mainly based on the theory of non-linear dynamic systems . This theory is supported by extensive modelling studies and evidence that vocal production is a highly non-linear dynamical system, in which changes caused by alterations in the vocal organs, muscles and nerves affect the dynamics of the whole system. Consequently, these changes can be detected by means of non-linear time series analysis tools or by means of computational approaches based on artificial intelligence. This project starts from the consideration that certain genetic abnormalities that cause a specific recognisable phenotype could also result in a specific vocal phenotype, or rather a 'phonotype'. Since vocal assessment is based on non-invasive and easily administered tests, vocal characterisation could be an informative tool in the diagnostic process and help both in defining the severity of clinical pictures and in performing genotype/phenotype correlations. Furthermore, voice studies could detect and monitor the progression of symptoms in certain genetic conditions that are often characterised by a regressive trend, such as in neuromuscular or metabolic syndromes. Smartphone technology, which has already been implemented in other fields, such as in dysphonia and Parkinson's disease, can be used to collect voice recordings of syndromic patients and can be an important tool to implement computerised assessment. Unlike most smartphone-based voice analysis tools, special attention will be paid to the reliability of audio recordings, both in the laboratory and at home, which will be performed according to a strict protocol. This will allow for uniform and reliable data and results. Artificial intelligence techniques will play a key role in studying the role of voice characterisation in diagnostic work for genetic syndromes. In addition, speech analysis could support the evaluation of the effectiveness of speech therapy, drug treatment and other rehabilitation approaches.
Primary Study Objective:
The primary objective of the study is the definition of distinct vocal phenotypes and the development of an Explained Decision Support System (DSS) for the automatic detection of vocal patterns in relation to the syndrome from which the patients suffer
Secondary:
1. Perceptual and acoustic analysis of voice recordings
2. Development of a voice recording collection system.
Number of participants: 500 (400 syndromic patients plus 100 non-syndromic controls matched by gender and age).
400 patients suffering from different syndromes will be recruited, although the investigators will focus in particular on Down syndrome, Noonan, Costello, Smith-Magenis, Cri du Chat, 22q11 deletion, Williams, Crisponi, Rubinstein Taybi and CHARGE to analyse their vocal pattern characteristics. The choice of the listed conditions is guided by their prevalence in the population and previously reported peculiar vocal patterns. The patients recruited are those regularly followed by the Centre for Rare Diseases and Congenital Defects of the Agostino Gemelli IRCCS University Polyclinic Foundation, Rome. Voice recordings collected in the laboratory from syndromic patients will be analysed both perceptually and objectively and compared with a control group of non-syndromic patients, matched for age and gender.
Recruitment & Eligibility
- Status
- NOT_YET_RECRUITING
- Sex
- All
- Target Recruitment
- 500
- Laboratory confirmation of the clinical diagnosis of one of the syndromes under investigation.
- Informed consent from parents, legal representatives or the patients themselves, according to the instructions of the hospital's Ethics Committee.
- Concomitant acute inflammatory disease of the upper respiratory tract.
- Refusal to sign the informed consent to participate in the study.
Study & Design
- Study Type
- INTERVENTIONAL
- Study Design
- PARALLEL
- Arm && Interventions
Group Intervention Description patient group voice recordings 500 (400 syndromic patients plus 100 non-syndromic controls matched for gender and age) control group voice recordings 500 (400 syndromic patients plus 100 non-syndromic controls matched for gender and age)
- Primary Outcome Measures
Name Time Method Vocal Phenotype definition of noise 2 years The definition of irregularity of voices of each syndrome will be extracted through PRAAT software.
PRAAT implements a method based on autocorrelation, applied to a time window of fixed size, and linear predictive coding. It requires the manual setting of some parameters. The software allows the analysis of noise (measured in dB).Vocal Phenotype definition of frequencies 2 years Distinct vocal phenotypes of each syndrome will be extracted through Biovoice software.
BioVoice allows the sequential analysis of several audio signals at once without any manual setting.
The software allows the analysis of fundamental and formants frequency (measured in Hz).Vocal Phenotype definition of irregularity 2 years The definition of irregularity of voices of each syndrome will be extracted through PRAAT software.
PRAAT implements a method based on autocorrelation, applied to a time window of fixed size, and linear predictive coding. It requires the manual setting of some parameters. The software allows the analysis of irregularity, namely jitter that is the relative average perturbation (measured in absolute jitter =1N-1∑i=1N-1\|Ti-Ti+1\|).
- Secondary Outcome Measures
Name Time Method Development of a system for collecting voice recordings - frequencies 2 years The development of a system to collect voice recordings will be made throught a Structured Query Language (SQL) database where all the voice recordings obtained will be stored in encrypted hard disks, accessible for data analysis and storage. For this project a cross-platform app, to record and manage the voice signal, for smartphones will be developed through an app framework (e.g Xamarin).
The following acoustic parameters will be stored and managed: fundamental and formants frequency (measured in Hz).Perceptual and acoustic analysis of voice recordings 2 years The perceptual evaluation of voice is aimed at further delineate the differences between syndromic and healthy subject voices. Judges will score the voice of each participant using the Grade, Instability, Roughness, Breathiness, Asthenia, and Strain (GIRBAS) scale.
GIRBAS scale provides a perceptual evaluation of voice according to five parameters: Grade (overall impression), Roughness, Breathiness, Asthenicity and Strain. Each parameter is rated on a 0 to 3 scale, where 0 indicates normality; 1, a slight deviance; 2, a moderate deviance; and 3, a severe deviance from normal.Other perceptual and acoustic analysis of voice recordings 2 years Hypernasality and hyponasality will be assessed using a 4-point ordinal rating scale with 1 representing normal resonance, 2, 3, and 4 representing mild, moderate and severe impairment.
Development of a system for collecting voice recordings - noise 2 years The development of a system to collect voice recordings will be made throught a Structured Query Language (SQL) database where all the voice recordings obtained will be stored in encrypted hard disks, accessible for data analysis and storage. For this project a cross-platform app, to record and manage the voice signal, for smartphones will be developed through an app framework (e.g Xamarin).
The following acoustic parameters will be stored and managed: noise (measured in dB).Development of a system for collecting voice recordings - irregularities 2 years The development of a system to collect voice recordings will be made throught a Structured Query Language (SQL) database where all the voice recordings obtained will be stored in encrypted hard disks, accessible for data analysis and storage. For this project a cross-platform app, to record and manage the voice signal, for smartphones will be developed through an app framework (e.g Xamarin).
The following acoustic parameters will be stored and managed: relative average perturbation (measured in absolute jitter).
Trial Locations
- Locations (1)
Fondazione Policlinico Universitario A. Gemelli Irccs
🇮🇹Roma, Italy