Diagnostic Algorithm for ADHD

Overseen By Beth Krone, PhD, MS

Age: < 18

Sex: Any

Trial Phase: Academic

Sponsor: MindTension

Must not be taking: Psychotropics

Disqualifiers: Seizure disorder, IQ<70, PTSD, others

No Placebo Group All trial participants will receive the active study treatment (no placebo)

Compensation: Varies

Number of Visits: 1

120 Participants Needed

Trial Summary

What is the purpose of this trial?

This study aims to demonstrate the accuracy of the MT1 algorithm using the MindTension biometric sensor device as a diagnostic aid for healthcare providers in diagnosing ADHD in youth ages ≥ 6 to ≤17 years.

Will I have to stop taking my current medications?

If you are currently taking stimulant medications, you will need to stop them for 3 days before testing. If you are on other psychotropic medications that can't be stopped in 3 days, you may not be eligible to participate.

What data supports the effectiveness of the treatment MT1 Auditory startle response patterns analysis algorithm, MT1 algorithm, MindTension biometric sensor device for ADHD?

The research suggests that wearable sensors and machine learning can effectively diagnose conditions like anxiety and depression in children, which implies potential for similar approaches in ADHD diagnosis. Additionally, machine learning models have shown high accuracy in predicting ADHD diagnoses, indicating that technology-based diagnostic tools, like the MT1 algorithm and MindTension device, could be effective. ^{1 2 3 4 5}

Is the MT1 algorithm safe for use in humans?

The research does not provide specific safety data for the MT1 algorithm or related devices in humans, but it mentions that the startle reflex, which the algorithm analyzes, is widely used in various studies, suggesting it is generally considered safe for research purposes. ^{6 7 8 9 10}

How does the MT1 algorithm treatment for ADHD differ from other treatments?

The MT1 algorithm treatment for ADHD is unique because it uses an auditory startle response pattern analysis, which is a novel approach compared to traditional methods that often rely on behavioral assessments or medication. This treatment involves a biometric sensor device that analyzes auditory responses, offering a potentially more objective and non-invasive way to diagnose and understand ADHD. ^{5 11 12 13 14}

Research Team

Jeffrey Newcorn, MD Professor

Principal Investigator

Director, Division of ADHD and Learning Disorders Icahn School of Medicine at Mount Sinai

Eligibility Criteria

This trial is for children aged 6 to 17 who may have ADHD. They must be willing to follow the study rules and can't have taken stimulants recently, or they must stop them for three days before testing. Kids with ADHD symptoms confirmed by specific assessments are included, while those without any diagnosable disorder on these tests are excluded.

Inclusion Criteria

1. Parent provision of signed and dated informed consent form

2. Child stated willingness to comply with all study procedures and availability for the duration of the study

3. Any gender, aged 6 to 17 years

Timeline

Screening

Participants are screened for eligibility to participate in the trial

1-2 weeks

Clinical Assessment

Participants undergo clinical assessment using the MT1 algorithm and standard diagnostic interviews

1-2 days

1 visit (in-person)

Follow-up

Participants are monitored for agreement between MT1 output and specialist clinician diagnosis

2 weeks

Treatment Details

Interventions

• MT1 Auditory startle response patterns analysis algorithm

Trial Overview The study is testing a new tool called the MT1 algorithm with a device named MindTension that measures how kids react to sounds. It's being compared to an established test (T.O.V.A.) to see if it helps doctors diagnose ADHD more accurately in young people.

Participant Groups

Group I: Non-ADHD group Experimental Treatment 2 Interventions

Subjects will be evaluated using clinically established reference standard interviews and psychometric tools with demonstrated validity, including the K-SADS and ADHD-RS-5, in accordance with DSM-5 criteria. IQ will be estimated using two subtests of the WASI.

Group II: ADHD group Experimental Treatment 2 Interventions

Subjects will be evaluated for ADHD using clinically established reference standard interviews and psychometric tools with demonstrated validity, including the K-SADS and ADHD-RS-5, in accordance with DSM-5 criteria. IQ will be estimated using two subtests of the WASI.

Who Is Running the Clinical Trial?

MindTension

Lead Sponsor

Trials

2

Recruited

160+

Findings from Research

A new 90-second fear induction task using wearable sensors can effectively diagnose anxiety and depression in young children, achieving an 80% diagnostic accuracy.

This method significantly reduces the time and cost associated with traditional diagnostic techniques, making it a promising tool for clinical settings in identifying internalizing disorders.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Rapid Anxiety and Depression Diagnosis in Young Children Enabled by Wearable Sensors and Machine Learning. McGinnis, RS., McGinnis, EW., Hruschak, J., et al.[2020]

The study highlights the potential of artificial intelligence (AI) to enhance the early diagnosis of ADHD by reviewing various diagnostic tools, including brain MRI, physiological signals, and performance tests, which could lead to improved treatment outcomes.

It identifies significant research gaps, such as the lack of publicly available datasets for ADHD assessment beyond MRI and the underutilization of data from wearable devices, suggesting that future work should focus on these areas to develop a comprehensive AI-supported diagnostic framework.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Automated detection of ADHD: Current trends and future perspective. Loh, HW., Ooi, CP., Barua, PD., et al.[2022]

The Wearable Diagnostic Assessment (WeDA) system demonstrated significantly higher sensitivity (94.55%) and specificity (98.18%) compared to the SNAP-IV scale (76.36% sensitivity and 80.36% specificity) in diagnosing ADHD among 55 children, indicating it may be a more accurate tool for diagnosis.

The WeDA system also achieved a higher area under the curve (AUC) of 0.964 compared to 0.907 for SNAP-IV, suggesting superior overall diagnostic performance, making it a promising option for clinical use.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

A Wearable Diagnostic Assessment System vs. SNAP-IV for the auxiliary diagnosis of ADHD: a diagnostic test. Luo, J., Huang, H., Wang, S., et al.[2022]

A multi-stage Bayesian classifier demonstrated high accuracy (over 86%) in predicting ADHD diagnoses based on parent and teacher surveys, suggesting that these low-cost measures can effectively supplement traditional diagnostic methods.

While the Bayesian approach performed well, it did not significantly outperform simpler methods like logistic regression or more complex ones like random forest, indicating that various classification techniques can be effective in ADHD diagnosis.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Prediction of Attention-Deficit/Hyperactivity Disorder Diagnosis Using Brief, Low-Cost Clinical Measures: A Competitive Model Evaluation. Mooney, MA., Neighbor, C., Karalunas, S., et al.[2023]

The traditional T-score method for diagnosing ADHD, which relies on a cut-off score of 70, has been found to have a high false negative error rate, making it less reliable for accurate diagnosis.

Novelty detection approaches, particularly methods like Parzen density estimators and one-class support vector machines, significantly improve classification accuracy for ADHD diagnosis, outperforming the T-score method by at least 30% for visual tests and 40% for auditory tests.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Supporting diagnosis of attention-deficit hyperactive disorder with novelty detection. Lee, HJ., Cho, S., Shin, MS.[2008]

A new machine learning algorithm has been developed to standardize the assessment of the startle reflex (SR) across various animal models, which can improve the reliability of data in psychophysiological research.

This standardized protocol aims to reduce variability in SR studies, making it easier to compare results across different laboratories and species, ultimately enhancing the understanding of conditions like schizophrenia and bipolar disorder.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Universal automated classification of the acoustic startle reflex using machine learning. Fawcett, TJ., Longenecker, RJ., Brunelle, DL., et al.[2023]

The study assessed the auditory startle reflex (ASR) in 27 healthy children, revealing that the blink response had a high response probability of 78%, while other muscle responses were significantly lower, indicating that the blink response is a key component of ASR in children.

Repetitive auditory stimuli led to a decrease in the combined response probability of multiple muscles and the sympathetic skin response, but the blink response remained stable, suggesting different habituation patterns for these responses in children.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Quantification of the auditory startle reflex in children. Bakker, MJ., Boer, F., van der Meer, JN., et al.[2014]

The startle reflex, originating in the caudal brainstem, can be measured through EMG recordings and consists of an initial motor reflex followed by a second response that includes emotional and voluntary behaviors.

Startle syndromes can be classified into three challenging groups: hyperekplexia, stimulus-induced disorders, and neuropsychiatric disorders, with neurophysiological investigations aiding in distinguishing between them and understanding their underlying causes.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Exaggerated startle reactions. Dreissen, YE., Bakker, MJ., Koelman, JH., et al.[2014]

The study successfully adapted three methodologies for measuring the startle response in 5-month-old infants, demonstrating their effectiveness in accurately recording motor reactions to acoustic stimuli.

All three methods—whole-body motor reactions, facial muscular contractions, and eyeblink intensity—showed consistent results in measuring the infants' startle responses, indicating reliable ways to assess this reflex in young children.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Three methodologies for measuring the acoustic startle response in early infancy. Agnoli, S., Franchin, L., Dondi, M.[2014]

The study measured the acoustic startle eyeblink response in 11 pediatric psychiatric inpatients, finding that children with ADHD had faster response latencies compared to those with other conditions, indicating heightened sensitivity to startle stimuli.

Using lower intensity sound stimuli (85-95 dB) effectively elicited startle responses in ADHD children, which may enhance testing compliance and comfort, allowing for longer and more effective assessment sessions.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Startle eyeblink elicitation in attention deficit disordered children using low-intensity acoustic stimuli. Goldstein, DJ., Blumenthal, TD.[2014]

The study analyzed resting-state EEG signals from 61 children with ADHD and 60 healthy children, revealing distinct EEG patterns that could aid in diagnosing ADHD.

Using advanced classification algorithms, particularly the Bernoulli Naive Bayes classifier, the research achieved a high accuracy of 96%, suggesting that EEG analysis could be a reliable supplementary tool for ADHD diagnosis.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Electroencephalogram (EEG) based prediction of attention deficit hyperactivity disorder (ADHD) using machine learning. Ahire, N., Awale, RN., Wagh, A.[2023]

The integrated device, combining EEG and CCD motion sensors, provides an objective measurement of ADHD by analyzing attention and movement patterns in preschoolers aged 3-5 years during structured tasks.

Statistical analysis showed a significant correlation between the device's objective assessment of ADHD and traditional clinical evaluations, suggesting it could enhance diagnostic accuracy for ADHD.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Integration of electroencephalogram (EEG) and motion tracking sensors for objective measure of attention-deficit hyperactivity disorder (MAHD) in pre-schoolers. Bhattacharyya, N., Singh, S., Banerjee, A., et al.[2022]

A study involving 181 adults with ADHD and 147 healthy controls demonstrated that machine-learning methods can effectively classify ADHD patients with a sensitivity of 75% to 83% and specificity of 71% to 77%, indicating a promising diagnostic tool.

The use of neuroalgorithms in diagnosing ADHD can improve reliability and objectivity by linking diagnostic outcomes to the underlying brain networks, although clinical implementation requires specialized facilities and expertise.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

EEG/ERP-based biomarker/neuroalgorithms in adults with ADHD: Development, reliability, and application in clinical practice. Müller, A., Vetsch, S., Pershin, I., et al.[2021]

A new method for diagnosing combined type ADHD using 24-hour actigraphy data shows high effectiveness, achieving 96.77% sensitivity and 84.38% specificity, making it a promising tool for accurate diagnosis.

This actigraphy-based approach is inexpensive and non-invasive, providing a practical alternative to more costly diagnostic methods, while also demonstrating that using features from the entire 24-hour period yields the best classification results.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Nonlinear analysis of actigraphic signals for the assessment of the attention-deficit/hyperactivity disorder (ADHD). Martín-Martínez, D., Casaseca-de-la-Higuera, P., Alberola-López, S., et al.[2012]

References

1. United States pubmed.ncbi.nlm.nih.gov

Rapid Anxiety and Depression Diagnosis in Young Children Enabled by Wearable Sensors and Machine Learning. [2020]

2. United States pubmed.ncbi.nlm.nih.gov

Automated detection of ADHD: Current trends and future perspective. [2022]

3. England pubmed.ncbi.nlm.nih.gov

A Wearable Diagnostic Assessment System vs. SNAP-IV for the auxiliary diagnosis of ADHD: a diagnostic test. [2022]

4. United States pubmed.ncbi.nlm.nih.gov

Prediction of Attention-Deficit/Hyperactivity Disorder Diagnosis Using Brief, Low-Cost Clinical Measures: A Competitive Model Evaluation. [2023]

5. Netherlands pubmed.ncbi.nlm.nih.gov

Supporting diagnosis of attention-deficit hyperactive disorder with novelty detection. [2008]

6. Netherlands pubmed.ncbi.nlm.nih.gov

Universal automated classification of the acoustic startle reflex using machine learning. [2023]

7. Netherlands pubmed.ncbi.nlm.nih.gov

Quantification of the auditory startle reflex in children. [2014]

8. Netherlands pubmed.ncbi.nlm.nih.gov

Exaggerated startle reactions. [2014]

9. United States pubmed.ncbi.nlm.nih.gov

Three methodologies for measuring the acoustic startle response in early infancy. [2014]

10. United States pubmed.ncbi.nlm.nih.gov

Startle eyeblink elicitation in attention deficit disordered children using low-intensity acoustic stimuli. [2014]

11. United States pubmed.ncbi.nlm.nih.gov

Electroencephalogram (EEG) based prediction of attention deficit hyperactivity disorder (ADHD) using machine learning. [2023]

12. United States pubmed.ncbi.nlm.nih.gov

Integration of electroencephalogram (EEG) and motion tracking sensors for objective measure of attention-deficit hyperactivity disorder (MAHD) in pre-schoolers. [2022]

13. England pubmed.ncbi.nlm.nih.gov

EEG/ERP-based biomarker/neuroalgorithms in adults with ADHD: Development, reliability, and application in clinical practice. [2021]

14. England pubmed.ncbi.nlm.nih.gov

Nonlinear analysis of actigraphic signals for the assessment of the attention-deficit/hyperactivity disorder (ADHD). [2012]