7. Biomarkers for objective measurement of ADHD?

Author: Ulrich Brennecke
Review: Dipl.-Psych. Waldemar Zdero (08/2024)

To date, there is no biomarker in sight that could be used to diagnose ADHD by measuring objective biological or neurological values.¹

However, measurement methods have already been researched extensively enough to validate and support diagnoses. The advantage of measurement methods is that the result is not distorted by prior knowledge, training or particular motivations (interest, hyperfocus).
However, it is known that persons with ADHD with high intrinsic interest can complete tests in a concentrated manner and with results indistinguishable from those of non-affected people. Therefore, tests in which the subjects are motivated to perform particularly well are falsified. It has also been reported that people with ADHD “outsource” the functions of disturbed brain areas to other areas of the brain in order to compensate for the deficit. Symptoms are also reduced through coping, for example by avoiding situations in which the stressful symptom can occur, or individual symptoms are reduced through intensive training. If only symptoms are asked about, the symptom masked by relocation to other areas of the brain or by coping may not be noticed. This is one of the reasons why tests alone can never be sufficient to reliably diagnose or rule out ADHD.
It is to be hoped that with further refinement of the measurement methods and knowledge about the evaluation of results, a more objective diagnosis than questionnaires and tests will be possible in the foreseeable future.

• 1. General information on biomarkers
  • 1.1. Types of biomarkers
  • 1.2. Combination of biomarkers improves results
  • 1.3. Biomarkers show snapshots
  • 1.4. AI test systems: major progress in the lab
    • 1.4.1. Detection rate development ADHD against non-affected persons
    • 1.4.2. Development of detection rates for ADHD versus other disorders
• 2. EEG / QEEG measurement
  • 2.1. Beta-theta ratio as a diagnostic tool
  • 2.2. EEG-Alpha-Power frontal
  • 2.3. Changed slope / offset of the 1/f noise in the EEG
  • 2.4. Measurement of evoked potentials
  • 2.5. Measurement of rsEEG
  • 2.6. Measurement of non-linear EEG values
  • 2.7. Phase space reconstruction of the EEG
  • 2.8. Gamma levels reduced in ADHD
  • 2.9. EPSPatNet86: 8-point EEG measurement on 85 wavelets
  • 2.10. EEG analysis using VMD-HT
  • 2.11. Brain complexity
  • 2.12. Dynamics between alpha, beta and gamma bursts
  • 2.13. Aperiodic exponent in EEG reduced
  • 2.14. Cortical inhibition at short intervals (SICI)
• 3. Functional near-infrared spectroscopy (fNIRS)
• 4. Genetic testing
  • 4.1. MiRNA
• 5. Endocrine and physical biomarkers
  • 5.1. Measurement of cortisol levels
    • 5.1.1. Measurement of the basal cortisol level
    • 5.1.2. Measurement of the phasic cortisol response to psychological stressors
  • 5.2. Dexamethasone/ACTH/CRH test?
  • 5.3. Mineral analysis
  • 5.4. Axonal damage to the white matter
  • 5.5. 24-Hour movement profiles
  • 5.6. Facial features
    • 5.6.1. Facial morphology in ADHD
    • 5.6.2. Facial morphology in ASD
  • 5.7. Analysis of MRI brain images
  • 5.8. Analysis of urine values
  • 5.9. Analysis of eye values
    • 5.9.1. Electroretinography (ERG)
    • 5.9.2. Pupillometry
    • 5.9.3. Eye tracker: duration of target fixation (?), duration of gaze on irrelevant areas
  • 5.10. Activity measurement / actigraphy
  • 5.11. Inflammatory biomarkers
    • 5.11.1. Platelet-lymphocyte ratio (PLR)
    • 5.11.2. Neutrophil to lymphocyte ratio (NLR)
    • 5.11.3. Monocyte/lymphocyte ratio (MLR)
    • 5.11.4. Mean platelet volume (MPV)
    • 5.11.5. Platelet distribution width (PDW)
    • 5.11.6. Plateletcrit (PCT)
    • 5.11.7. Leukocyte count (MBC)
    • 5.11.8. Basophil count
    • 5.11.9. Eosinophil count
    • 5.11.10. Anti-myelin proteins
    • 5.11.11. Neutrophil-to-high-density lipoprotein ratio
    • 5.11.12. Lymphocyte-to-high-density lipoprotein ratio
    • 5.11.13. Monocyte-to-high-density lipoprotein ratio
    • 5.11.14. Platelet-to-high-density lipoprotein ratio
    • 5.11.15. CRP
    • 5.11.16. Procalcitonin
    • 5.11.17. Systemic immune inflammation index
  • 5.12. Intrinsically disordered regions of active proteins (IDR)
  • 5.13. Metabolite profiles
  • 5.14. Language analysis
    • 5.14.1. Analysis of voice recordings
    • 5.14.2. Stylometric features

1. General information on biomarkers¶

1.1. Types of biomarkers¶

A distinction is made between the following types of biomarkers:²

• Diagnostic biomarkers: predict the presence of a disease; support diagnosis
• Predictive biomarkers: predict the most likely response to a certain type of treatment
• Prognostic biomarkers: indicate disease progression over time with or without treatment
• Mechanistic biomarkers: indicate the underlying pathophysiological and/or psychological process
• Surrogate outcome biomarkers: serve as surrogate endpoints for a relevant clinical outcome to be observed later
• Stratification biomarkers: divide heterogeneous diseases into more homogeneous groups
• Genetic markers: identify changes in DNA or RNA (such as mutations)
• Molecular markers: proteins, hormones or other molecules in the blood or other body fluids
• Physiological markers: body temperature, blood pressure, respiratory rate
• Imaging markers: Visible changes that can be detected by X-ray or MRI

1.2. Combination of biomarkers improves results¶

We suspect that, as with symptom testing, there will be no single biomarker that can diagnose ADHD in individuals with sufficient certainty. ADHD is a syndrome, which means that symptoms can arise from many different causes.
From the perspective of a syndrome, biomarkers are much closer to the neurophysiological causes than behavioral symptoms. Therefore, finding a single unique biomarker that clearly identifies ADHD is likely to be even less likely than finding a single symptom.
We therefore assume that there will be no single biomarker for ADHD that would be sufficiently diagnostically valid, but that a multidimensional approach can achieve good diagnostic quality. When diagnosing on the basis of symptoms, it is also for good reason that the focus is not on individual symptoms, but on a certain number of symptoms from a symptom group.
Studies show that a combination of several biomarkers improves diagnostic accuracy.³

• A study in young children with ADHD using a portable, wireless EEG meter with machine learning achieved 95% accuracy (2024). Accuracy increased to 97.4% when two additional markers were integrated (computerized attention assessment using Conners’ Kiddie Continuous Performance Test Second Edition (K-CPT-2) and ratings from ADHD-related symptom scales).⁴
• A combination of 4 biomarkers (increased HRV, decreased EDA, pulse arrival time and respiratory rate) achieved an accuracy rate of 85.5% and an AUC value of 0.95 when adjusting for gender and age in differentiating children with ADHD from healthy children.⁵
• A study with three biomarkers (HRV, electrodermal activity, skin temperature) achieved a sensitivity of 81.4 % and a specificity of 81.9 %. Here, too, the integration of data from all physiological signals delivered the best results⁶
• A combination of 15 blood values from children and adolescents with ADHD achieved an AUC of 87 % with regard to the detection of ADHD, while each of the 15 blood values alone only achieved an AUC of between 54 and 68 %.⁷ The model can be used experimentally with your own blood values at https://adhdrisk.streamlit.app. ADHD diagnosis-relevant were:
  • 25 Hydroxyvitamin D
  • Absolute eosinophil count
  • Aspartate aminotransferase
  • Calcium
  • Carbon dioxide
  • Direct bilirubin
  • Lactate dehydrogenase
  • Magnesium
  • Percentage of eosinophils
  • Phosphonium
  • Prealbumin
  • Total bile acids
  • Total bilirubin
  • Urea
  • Uric acid
  • Distribution width of the red blood cells
  • Beta-2-microglobulin in the blood
• Multimarker study:⁸
  • Pupil characteristics alone: accuracy of 84.4 %, AUROC 88.6 %
  • Pupil characteristics and task performance together: Accuracy 86.7%, AUROC 91.5%
  • Pupil characteristics, task performance and reaction time metrics: Accuracy of 88.9 %, AUROC 90.8 %
    • Sensitivity 97.8 %
    • Specificity 82.2 %

Such values sound good, but practical use does not require the detection of a disorder in an environment of healthy controls, but diagnostic accuracy in a natural population group, in which not only Disorder X must be differentiated from healthy controls, but also from all other disorders occurring in the population. This places considerably higher demands.

1.3. Biomarkers show snapshots¶

A fundamental problem with measuring biomarkers is that they can only ever provide a snapshot. The results of this snapshot also depend on whether the person with ADHD is experiencing acute stress at the time of measurement or not. The usual ADHD diagnosis is therefore correctly based on whether the (stress/ADHD) symptoms exist in the long term, i.e. outside of acute circumstances that trigger particular stress.
To date, the relevant literature has not yet addressed the extent to which biomarkers can reliably distinguish long-term changes from merely acute or chronic stress reactions.

1.4. AI test systems: major progress in the lab¶

Since 2023, AI knowledge systems such as chat GPT have been on everyone’s lips. Machine learning or deep learning are methods for using AI methods

Compared to the varying reliability of mental health diagnoses, the research results in the laboratory already show very impressive values.⁹
Nevertheless, the results of such test systems are a long way from actually being usable. The detection rates usually only refer to small databases of 20 to a few hundred data sets (given in n), consisting exclusively of data from people with ADHD and from (known) healthy subjects (-> 2.1.). In practice, however, ADHD can also be distinguished from an unknown health status into a mixture of healthy individuals and persons with other mental disorders. Under 2.2. we have collected the first studies that always distinguish between individual predefined disorders.

1.4.1. Detection rate development ADHD against non-affected persons¶

The temporal and qualitative development of the detection rates gives a good impression of the potential of AI diagnostics. The % figures indicate the accuracy, the year figures indicate the publication date of the descriptive publication.

• 100 % (2025) by machine learning using a minimum spanning tree graph based on similarities of EEG data of different brain areas in different EEG frequency subbands. The most discriminative graph features were found in the alpha band: a significantly reduced number of leaves and the average values of eccentricity, radius and diameter in the high alpha range showed a lack of frontal processing centers and a weaker frontoparietal connection in the ADHD group.¹¹
• 99.58% (2025) by machine learning with the CatBoost algorithm and feature importance scoring with the SHAP algorithm using multidimensional EEG features in children with ADHD compared to healthy controls. ADHD correlated with an increase in the power of theta, alpha and beta rhythms, with increased power ratio between theta and beta (theta/beta ratio, TBR) and with a reorganization of whole brain functional connectivity across all frequency bands (here primarily increased functional connectivity.¹²
• 99.42% accuracy, 99.03% precision, 99.82% recall, 99.42% F1 score (2025) by EEG signal analysis via deep learning using a ResNet-based model with a doubly extended attention mechanism consisting of autoencoder for feature extraction, Reptile Search Algorithm for feature selection and modified ResNet architecture for model training.¹⁰
• 99.2 % (2025) by analyzing simple video recordings of test subjects in a computer test situation¹³
• 99.17 % (2024) in a data set of children with ADHD and healthy children using EEG data from the theta band of the frontal and occipital lobe (n = 121)¹⁴
• 98.88 % (2024) through a combination of linear and non-linear maps of brain connectivity with an attention-based convolutional neural network (Att-CNN)¹⁵
• 98.86 % (2024) through Skip-Vote-Net, a deep learning-based network¹⁶
• 98.77 % (2023) through a new CNN based on fMRI data sets¹⁷
• 98.7% (2024) using video data with facial, postural and hand features of adults with ADHD through machine learning (n = 22)¹⁸
  • Accuracy of 98.67 %
  • Precision of 98.01 %
  • Recognition of 98.88 %
• 98.52% (2025) by analyzing EEG, eye-tracking or surveys to detect patterns associated with ADHD symptoms using deep learning using a RhinoFish optimization algorithm to select optimal features (F-score 98.26%, specificity 98.16%)¹⁹
• 96.9 % AUC (2025) through machine learning from images of the retina²⁰
• 96.5 % (2024) using the hybrid PUDMO algorithm for the evaluation of EEG data²¹
• 96 % (2024) by the convolutional neural network CNN model RBP, which was originally developed for the precise classification of brain tumors in medical imaging (n = 3000)²²
• 95.3% (2025) by machine learning using the Isolation Forest algorithm and subsequently a recurrent neural network (RNN) model, using the HYPERAKTIV dataset, which contains high-resolution temporal data on the motor activity of people diagnosed with ADHD.²³
• 95% accuracy (2024) in young children with ADHD using a portable wireless EEG meter with machine learning. Accuracy increased to 97.4% when two additional markers were integrated (computerized attention assessment using Conners’ Kiddie Continuous Performance Test Second Edition (K-CPT-2) and ratings from ADHD-related symptom scales).⁴
• 94.5 % (2024) with the LRP algorithm (Layer-wise Relevance Propagation) based on EEG data²⁴
• 93.9 % (2024) Frequency-Integrated Visual-Language Network (FIVLNet), a deep learning framework for analyzing MRI scans²⁵
• 92 % (2025) Machine learning model of analyzing written natural language reports from patients who had undergone intelligence testing. Enabled the reconstruction of human-readable text from the selected N-gram features.²⁶
• 90.81 % (2025) Machine learning from EEG data of n = 2434 19-channel EEG data recordings from 5 frequency bands of n = 168 subjects²⁷
• 90.02 % (2021) accuracy, AUC 95.7 %, sensitivity 89.8 % and specificity 90.55 % achieved an analysis of ADHD gene candidates using deep learning (“AI”) by combining genes with insignificant P-values.²⁸
• 90 % (2024) random forest algorithm; AUC 94 %, sensitivity 91 %, specificity 92 %²⁹
• 89.89% (2025) by near-infrared spectroscopy (fNIRS) during verbal tasks in conjunction with machine learning (ML) to discriminate between adults with ADHD (n = 120) and healthy controls (n = 75)³⁰
• 89 % (2025) Random forest model based on EEG data³¹
• 88.6 % (2024) Random forest model for resting-state theta-band EEG data³²
• 88 % (2020) using a convolutional neural network (CNN) to analyze EEG data, in particular event-related potentials collected during the flanker task³³
• 87 % (2020) Continuous performance test evaluated with machine-based learning model. Sensitivity 89 %, specificity 84 %³⁴
• 86.4 % accuracy and 95.5 % sensitivity (2025) through machine learning on EEG signals from auditory event-related potentials³⁵
• 82 % (2023) accuracy for a special AI system for the diagnosis of ADHD. The sensitivity was 64 % and the specificity 99 %. The accuracy was higher than for all other disorders tested with this system³⁶
• 79 % (2023) through a combination of 4 machine learning algorithms for analyzing register data.³⁷
• 75.6% (2021) using a combination of Convolutional Denoising Autoencoders (CDAE) classification method and Random Forest (RF) algorithms, with a sensitivity of 76.9% and a specificity of 73.1%³⁸
• 74.9% (2020) by a multimodal 3D-CNN using gray matter features and fALFF from fMRI³⁹
• 69.15 % (2017) using a deep learning model to classify MRI scans⁴⁰

1.4.2. Development of detection rates for ADHD versus other disorders¶

It is much more difficult to distinguish between different disorders than between ADHD and healthy people.
Systems will only be suitable for practical use when they are able to recognize disorders in a large, non-predefined group of test subjects. There is still a very long way to go until then, not unlike automated driving, which, even after many years of development, still only works in known environments or very restricted environments (highways)

• 85% accuracy (2025) in distinguishing between ADHD and ASD⁴¹,
• 87% accuracy (2025) of detection of ASD in males versus healthy individuals, 84% for ADHD in females versus healthy individuals, 70% in the 3-group model (ASD, AuDHD, ADHD) and 53% when all groups were included (ASD, ADHD, AuDHD and controls) using electroretinograms with machine learning.⁴²

Further AI systems for diagnosing ADHD are currently under development.⁴³

2. EEG / QEEG measurement¶

The surface EEG signal (spontaneous electroencephalography) is the image of all electrical events of the entire head and includes action potentials, postsynaptic potentials and the electrical signals of the skin, muscles, blood and eyes. Different EEG waves correlate with certain states of consciousness. Some cerebral pathologies (e.g. epilepsy) have specific EEG patterns. The amplitude of EEG signals decreases with increasing propagation from their source. Therefore, multi-channel measurement of the EEG signals at different points on the scalp allows the origin of certain waves to be identified.⁴⁴

The measurement of QEEG or EEG can support diagnoses. Further information on QEEG and neurofeedback therapy for ADHD can be found at ⇒ Neurofeedback as ADHD therapy In the chapter Treatment and therapy.

A multifactorial 19-channel EEG analysis called “ADHD-AID” reported an ADHD diagnostic accuracy of:⁴⁵

• Accuracy: 0.991
• Sensitivity: 0.989
• Specificity: 0.992
• F1 score: 0.989
• Mathew correlation coefficient: 0.982.
• AUC: 0.9958

Studies reported an ADHD detection rate of an EEG transformer model with an average accuracy of 95.85% and an average AUC value of 0.9926.⁴⁶ An EEG classification model with machine learning reported a classification accuracy of 98.28% to 98.86%.⁴⁷ Using phase-based analysis, one study identified two biomarkers that distinguished ADHD in children from healthy children with an accuracy of 99.174%:⁴⁸

• the subgraph centrality of the phase shift index of brain connectivity within the beta and delta frequency bands
• the node betweenness centrality of the inter-site phase cluster connectivity within the delta and theta bands

In practice, however, people with ADHD do not have to be distinguished from healthy people, but ADHD has to be distinguished from other disorders of unknown composition. This is a much more complex task than all the methods presented so far have been able to solve.

2.1. Beta-theta ratio as a diagnostic tool¶

Attempts have been made to use the theta-beta ratio as a diagnostic tool for ADHD.
Although individual studies on this have appeared promising, this diagnostic approach does not yet seem suitable for diagnosing ADHD.

The theta-beta ratio is said to be significantly altered in ADHD.

• Increased power in delta, theta and low-alpha over central parietal regions
• Increased power in low beta in frontal regions
• Increased power in mid-beta over parietal regions.

Our discussions with a neurofeedback therapist revealed some interesting findings:

• Not only in ADHD, but also in a number of other mental disorders, such as depression, Compulsions, trauma or even migraine sufferers, the beta increases in a state of relaxation, while it decreases in healthy people when they relax.
  This - characteristic - increase in beta levels during relaxation could possibly explain the symptoms of racing thoughts and sleep problems.
  We conclude from this that the theta-beta ratio in the laboratory environment can at best enable a clear diagnosis to be made as a differentiation of a group of people with ADHD in comparison to a group of (proven) healthy people. However, it is probably not suitable as a diagnostic tool in real life - for people for whom it is not known whether they suffer from ADHD or another Disorder, or are healthy - because deviations in the theta and beta range occur in a variety of mental disorders and the alternative to the presence of ADHD, unlike in the test environment, is not necessarily “healthy”, but includes the possibility of a number of other mental disorders or tendencies.
• ADHD-HI and ADHD-I have characteristically different theta/beta profiles.
  • In ADHD-HI and ADHD-C (with hyperactivity), theta is typically too low and beta is typically too high
  • In people with ADHD-I (without hyperactivity, dreamers), theta is typically too high and beta is typically too low. Here, the theta-beta ratio should therefore be too low - contrary to the postulation in the studies cited above.
  • In both groups, theta/beta training (relaxation training) proves to be helpful - in ADHD-HI, theta is trained up and beta is trained down at the same time, while in ADHD-I it is the other way round, reducing theta and increasing beta (concentration training). In both cases, it is not only the objective change in the values that is therapeutically effective, but also the learned ability to influence the values, even if this results in the values changing towards mean values.

Joint tests with a neurofeedback therapist revealed a significant influence of medication and food (sugar) on the QEEG values.

• Methylphenidate causes greater agitation within the QEEG values within a few minutes. As expected, methylphenidate impeded the theta increase and beta decrease in ADHD-HI targeted by neurofeedback. This is not surprising as MPH is a stimulant. It explains why MPH should be taken in such a way that the effect has worn off by bedtime.
• Sugar (2 chocolate bars within 5 minutes at 90 kg body weight) caused significant changes in the QEEG of an ADHD-HI subject.
  • Beta1 rose considerably within 10 minutes.
    The threshold values for theta up / beta down training had to be reduced considerably. The 85 % of the target values achieved before the sugar intake had dropped to 50 %. This means that the ability to relax had decreased drastically.
  • After 20 minutes, all values (theta, alpha, beta1, beta 2, hi-beta) were significantly reduced. Relatively speaking, however, beta 1 was now clearly above SMR. (Beta1 should be below SMR, which is why SMR training (which targets it) is the first step in neurofeedback treatment for ADHD).
  • After 30 minutes, Beta1 had caught up slightly with SMR. However, Hi-Beta was now significantly higher.

There is also the problem that different EEG measurement software products apparently use different calculation methods to determine the beta-theta ratio and therefore produce significantly different results with the same data.⁵⁶ Therefore, the International Collaborative ADHD Neurofeedback (ICAN) randomized clinical trial used a fixed theta-beta ration cutoff of ≥ 4.5, which had to be measured with the Thought Technology Monastra-Lubar Assessment Suite, 1.5 SD above the norms collected with this system.

• Binaural music also has a significant influence on the change in the QEEG.
  ⇒ Binaural music for ADHD and sleep problems.

One study came to the conclusion that the theta-beta ratio of the EEG in both non-affected and persons with ADHD-C is a reflection of cognitive processing activity and can be measured using the P 300 latency in the acoustic oddball test. In contrast, absolute alpha power did not correlate with P 300 latency or amplitude in persons with ADHD-C.⁵⁷

2.2. EEG-Alpha-Power frontal¶

In 5-month-old children who were later diagnosed with ADHD, there was reduced mean frontal EEG activity during an attention task. Since only the group values differed, diagnostic use was not possible.
⁵⁸

2.3. Changed slope / offset of the 1/f noise in the EEG¶

“1/f noise” in the EEG are arrhythmic signals in the cortex that are typical of ADHD as neural noise. Increased individual reaction time variance is a sign of increased neural noise. MPH improves this.⁵⁹

In 3- to 7-year-old children with ADHD, a steeper slope and a larger offset were found with regard to the “1/f noise” in the EEG. These data correlated with altered theta/beta ratios. Both values were normalized by medication. In addition, a greater alpha power was found.⁶⁰

2.4. Measurement of evoked potentials¶

Event-related potentials are waveforms in the electroencephalogram (EEG) that are related (correlated) to an observable event (sensory stimulus, movement).
Evoked potentials are event-related potentials in which a test subject always receives the same stimulus without further instruction, so that it is purely perceptual.

The measurement of evoked potentials is an objectified diagnostic procedure. Characteristic EEG amplitude curves triggered by stimuli are measured in various regions of the brain typically affected by ADHD. A typical average amplitude curve is determined from a large number of test runs in order to determine the typical signal change caused by the event and to separate it from the background noise. These EEG traces can be compared with data from non-affected people and data from people with other disorders, which are collected in so-called QEEG databases.
Unfortunately, diagnosis using evoked potentials is time-consuming (4 hours of tests, which are subsequently evaluated) and therefore not cheap. At the Brain and Trauma Foundation Graubünden Switzerland in Chur, a diagnosis based on evoked potentials costs around CHF 1,250 (as of 2015).

The diagnostic accuracy of measuring evoked potentials is growing rapidly. While Strehl et al. cite studies from 2005 with a (still unsatisfactory) diagnostic accuracy of 70 to 80 %, Müller et al. 2011⁶¹ already claim 89 % diagnostic accuracy: from a group of 212 adults with 106 people with ADHD, 89 % of the persons with ADHD were correctly diagnosed using a 19-channel system by automatically evaluating EVP. Models with machine learning achieve 94 %⁶²
However, these results were determined in laboratory situations in which only ADHD could be distinguished from those not affected. In real-life diagnostics, ADHD can be distinguished from a variety of other disorders. It remains to be seen when this will be possible in practice using such models.

By comparison, trained diagnosticians and experienced clinicians achieved a diagnostic agreement of 88%⁶³

One study found no significant differences in evoked potentials between twins with and without ADHD in a continuous performance task.⁶⁴

A meta-analysis of 52 studies with 3,370 subjects on early (P100, N100, P200, N200, ERNNe) and late (P300, Pe, CNV) evoked potentials found in ADHD⁶⁵

• Shorter Go-P100 latencies
• Smaller Cue-P300 amplitudes
• Longer Go-P300 latencies
• Smaller NoGo P300 amplitudes
• Longer NoGo P300 latencies
• Smaller CNV amplitudes
• Smaller Pe amplitudes.

with considerable heterogeneity of the results and moderate effect sizes (d<0.6).

Another study came to the conclusion that evoked potentials show very different correlations with symptoms depending on the age of the test subjects (childhood or adolescence).⁶⁶

The mismatch negativity (MMN) is a negative component of the event-related response in an EEG signal that is triggered by a perceptible change in a repetitive aspect of auditory stimulation (e.g. stimulus pitch, stimulus duration).⁶⁷
Children with ADHD showed reduced mismatch negativity (MMN) amplitudes in response to two blocks of duration- and ISI-based deviants:⁶⁸

• MMN amplitude measurements:
  • Fz
    • ADHD: -1.2097 ± 0.2938 (continuous block); -0.8553 ± 0.4423 (ISI block)
    • Controls: -1.8325 ± 0.3689 (continuous block) and -2.0855 ± 0.3802 (ISI block)
  • Cz
    • ADHD: -1.2515 ± 0.3261 (continuous block) and -0.9367 ± 0.3432
    • Controls: -2.1319 ± 0.4445 (continuous block) and -2.7561 ± 0.4883
• MMN latencies
  Children with ADHD showed longer MMN latencies in both experimental blocks, indicating atypical responses.
  • Fz:
  • ADHD: 239.68 ± 5.059 (continuous block) or 226.88 ± 4.885 (ISI block)
  • Controls: 228.56 ± 6.584 (continuous block) or 213.56 ± 4.153 (ISI block)
  • Cz
    • ADHD: 234.40 ± 5.741 (continuous block) and 231.44 ± 5.464 (ISI block)
    • Controls: 227.52 ± 6.710 (continuous block) and 218.00 ± 5.261 (ISI block)

2.5. Measurement of rsEEG¶

The measurement of 14 resting-state EEG parameters was able to identify the participating people with ADHD with an accuracy of around 85% in a process optimized by machine learning.⁶⁹ As it is not only possible to diagnose healthy people in the wild, but also other disorders, it is questionable whether the result will be accurate enough to be useful in practice.

2.6. Measurement of non-linear EEG values¶

According to a study, a measurement of non-linear EEG values showed a better diagnostic quality of ADHD than other EEG diagnostic methods.⁷⁰

2.7. Phase space reconstruction of the EEG¶

One study found that phase space reconstruction of the EEG is very good at distinguishing ADHD from non-affected individuals. ⁷¹ For diagnostic use, however, a method must also be able to differentiate well between other disorders.

2.8. Gamma levels reduced in ADHD¶

One study reported elevated resting gamma levels in adults with ADHD in the 30 to 39 Hz range (gamma1), but not in the 39 - 48 Hz range (gamma2). Resting gamma1 power increased with age and was significantly lower in ADHD than in control subjects in early adulthood.⁷²

2.9. EPSPatNet86: 8-point EEG measurement on 85 wavelets¶

A study of 85 EEG wavelet factors was able to correctly classify 87% of people with ADHD using machine learning.⁷³ It remains to be seen what the recognition rate will be for an open group of test subjects.

2.10. EEG analysis using VMD-HT¶

One study used VMD-HT and other secondary EEG characteristics to distinguish between 61 people with ADHD and 60 healthy controls:⁷⁴

• an accuracy of 99.81 %
• a sensitivity of 99.78 %
• a specificity of 99.84 %
• an F-1 measure of 99.83 %
• a precision of 99.87 %
• a false detection rate of 0.13 %.
• Detection rate for ADHD of 99.87 %
• Detection rate for healthy controls of 99.73 %

These impressive figures will have to be proven in the real world when not only persons with ADHD and healthy controls, but also people with other disorders are tested.

2.11. Brain complexity¶

An analysis of brain complexity in ADHD, including using EEG, found that brain complexity is increased in children with ADHD (with the increase being weaker than in schizophrenia (positive traits) and significantly weaker than in depression) and decreased in adults with ADHD (with the decrease being weaker than in autism and significantly weaker than in dementia).⁷⁵

2.12. Dynamics between alpha, beta and gamma bursts¶

The dynamics between alpha/beta and gamma bursts in the dlPFC and posterior parietal cortex during a visual Sternberg task on spatial working memory showed a complementary role to working memory processes in adolescents with ADHD. Alpha/beta bursts decreased during encoding of the stimuli and increased during the delay, whereas gamma bursts were increased during encoding and decreased during the delay. Deviations in bursting patterns were associated with working memory errors and clinical symptoms.⁷⁶

2.13. Aperiodic exponent in EEG reduced¶

The aperiodic exponent and aperiodic dynamics are indicators of age-related maturation of the cerebral cortex and the E:I ratio and are clearly related to ADHD symptoms and cognitive abilities.
The aperiodic resting electroencephalographic (EEG) activity is dynamic and reflects the changing balance between excitation and inhibition (E:I) under changing environmental conditions. Resting EEG on n = 285 children aged 2 to 14 years under high and low visual stimulus exposure showed⁷⁷

• a linear increase in the aperiodic exponent from early to middle childhood
  • –> low E:I value, followed by declining development in late childhood.
• The aperiodic exponent was
  • greater in younger children with high visual stimulation than with low visual stimulation
  • this effect was reversed with increasing age
• ADHD correlated with a reduced aperiodic exponent
• IQ correlated with dynamic aperiodic activity, i.e. shifts in the E:I equilibrium

2.14. Cortical inhibition at short intervals (SICI)¶

Decreased short-interval cortical inhibition (SICI) is a consistent finding in children with ADHD. Paired-pulse SICI showed good test-retest reliability (ICC > 0.75), whereas intracortical facilitation (ICF) showed only moderate test-retest reliability (ICC > 0.50).⁷⁸

3. Functional near-infrared spectroscopy (fNIRS)¶

A prospective study of 30 boys using fNIRS of the forehead while performing mental arithmetic tasks was able to identify almost 100 % of the 15 boys with ADHD.⁷⁹ So far, this is a purely experimental approach.
Another study found no differences by fNIRS in ADHD⁸⁰

4. Genetic testing¶

To date, there are no usable genetic tests for the diagnosis of ADHD.
However, initial voices are now (2020) in favor of integrating certain genetic tests (here: array CGH) into the standards of ADHD diagnostics.⁸¹

There are several difficulties for genetic diagnostics in ADHD:

• ADHD is usually not caused by individual gene variants, but by the interaction of a large number (several hundred to thousands) of gene variants (⇒ gene candidates in ADHD) Gene candidates for ADHD)
• The (epigenetic) expression of genes (e.g. through early childhood or chronic severe stress) can also change the gene effect independently of the gene variant.
• It can be assumed that different combinations of the genes and forms of expression under consideration coincide in different persons with ADHD, but not all genes that contribute to the dopamine balance being disturbed, for example.
• There is not yet a complete set of candidate genes, which is why there is no genetic test for ADHD. We have collected over one hundred and fifty gene candidates.
  ⇒ Gene candidates for ADHD.
• Genetic tests are still too expensive to be able to test the required number of genes.
  It is possible that genetic tests will become so affordable in the coming years that they can also be used for diagnostic purposes - and subsequently to adapt individualized treatment.

4.1. MiRNA¶

One study investigated miRNAs of extracellular vesicles in serum. A longitudinal analysis examining changes in miRNA expression over time between four groups with different diagnostic trajectories (persistent diagnosis, first onset, remitted and typical development/control) found no statistically significant results. In the cross-sectional analysis, miR-328-3p was found to be upregulated in only one of 2 ADHD groups. These miRNAs may regulate the expression of genes associated with these traits in genome-wide association studies⁸²

One review found in children with ADHD⁸³

• highly regulated
  • miRNA-let-7d
  • miR-132-3p
  • miR-5692b
• down-regulated
  • miR34c-3p
  • miR-138-1
  • miR-107
  • miR-142-3p
  • miR-378
  • miR-30e-5p
  • miR-140-3
  • miR-1265p

Treatment success in ADHD by MPH (responders) at 12 months correlated with increased expression of miR-140-3p, miR-27a-3p, miR-486-5p and miR-151-5p.⁸⁴

5. Endocrine and physical biomarkers¶

5.1. Measurement of cortisol levels¶

The cortisol value has two meanings:

• The basal = tonic = long-term value represents the basic level, which changes rhythmically throughout the day (circadian rhythm)
• The phasic = short-term value represents a response reaction to an acute stressor

5.1.1. Measurement of the basal cortisol level¶

The basal cortisol level is lower in people with ADHD than in those not affected because the stress load is long-term. In the case of short-term stress, the basal cortisol level initially rises.

This difference could theoretically be used to differentiate between an acute stress reaction (increased cortisol levels) and a collapsed HPA axis regulation (reduced levels) - if the individual fluctuations between people within a group were not greater than the increase or decrease in cortisol levels compared to healthy levels.

According to this hypothesis, however, an individual change in cortisol levels upwards or downwards could be determined if these were recorded regularly (annually) from the earliest age (in children: on the occasion of standard examinations).
Without historical values of the basal cortisol level, however, such a measurement is useless.

In addition, a measurement of the morning cortisol awakening response (CAR) could provide information about the state of the HPA axis. The CAR is a persistent increase in cortisol levels that occurs 20 to 40 minutes after waking up.

A comparative graphical representation of the cortisol profiles of healthy, acutely stressed, chronically stressed, partially burned out and fully burned out people can be found in Bieger.⁸⁵

However, here too there is the difficulty that without individual comparative data, an assessment of the 24-hour picture of cortisol values is barely diagnostically usable.

A measurement of cortisol in hair (which reflects long-term levels) showed that low cortisol levels in preschool children predicted the expression of ADHD.⁸⁶
This could possibly be the result of a long-term stress reaction. The third phase of stress development (resistance phase) is typically characterized by a collapse of basal cortisol levels. The fourth phase (exhaustion phase) is then characterized by a breakdown of the neurotransmitter systems. More on this at ⇒ Resistance phase In the section ⇒ The stress response chain / stress phases of the article ⇒ The human stress system - the basics of stress.

However, changes in basal cortisol levels are not specific to ADHD and are therefore barely useful for ADHD diagnosis.

5.1.2. Measurement of the phasic cortisol response to psychological stressors¶

The phasic cortisol response can be used to determine the cortisolergic subtype of ADHD.

ADHD-I is typically associated with an elevated cortisol response compared to unaffected individuals, ADHD-HI with a flattened cortisol response compared to unaffected individuals.

The differences are relevant for treatment. In people with ADHD-I, SSRIs should be used with caution. In ADHD-HI, low-dose SSRIs (2 to 4 mg, i.e. 1/5 to 1/10 of the level used in depression) could potentially improve impulsivity, allowing lower dosing with stimulants.

The Trier Stress Test (TSST) is typically used as a psychological stressor.
This is probably not suitable for a (even annually) repeated examination in view of the expected habituation.

5.2. Dexamethasone/ACTH/CRH test?¶

A dexamethasone/ACTH/CRH test can be used to determine whether there is permanent dysregulation at individual levels of the HPA axis.

If ADHD is characterized by a permanent dysregulation of the stress regulation systems, above all the HPA axis, it should be possible to prove the individual dysregulation by testing the HPA axis reactions (changes in the levels of the stress hormones CRH, ACTH, cortisol to acute stressors). At the very least, such tests should provide informative indications for the individual differentiation of subtypes.

Dexamethasone is a glucocorticoid that selectively binds to the glucocorticoid receptors and should thereby trigger the downregulation of the HPA axis (after a stress reaction has occurred). If this downregulation does not occur, this is a strong indication of a disorder in the shutdown of the HPA axis.

More on this at ⇒ Pharmacological endocrine function tests.

5.3. Mineral analysis¶

By analyzing zinc, lead, copper, cobalt and vanadium from teeth, one study was able to create an impressively clear distinction between non-affected people, people with ADHD, people with ASD and people with comorbidities.⁸⁷

5.4. Axonal damage to the white matter¶

A study of 50 untreated persons with ADHD compared to their unaffected twins and 50 control subjects found higher axonal diffusivity in the ADHD sufferers in the⁸⁸

• Perpendicular fasciculus
• Superior longitudinal fasciculus I
• Lateral corticospinal tract
• Corpus callosum

The values correlated significantly (except in the perpendicular fasciculus) with ADHD symptoms such as inattention and working memory problems and were intermediate between those of people with ADHD and non-affected twins. Axial diffusivity is a marker for the undamaged condition of axons.

5.5. 24-Hour movement profiles¶

One study claims to have achieved a diagnostic accuracy for ADHD-C of over 97% by evaluating neural network analyses of 24-hour movement profiles.⁸⁹⁹⁰

5.6. Facial features¶

5.6.1. Facial morphology in ADHD¶

One study concluded that there may be a close relationship between ADHD and nasal width, ear length and upper face depth, which could be related to genetic signaling processes and a close relationship between the brain and face formation process in the embryonic period.⁹¹

5.6.2. Facial morphology in ASD¶

A study reports on subtle phenotypic facial morphologies in ASD.⁹²

5.7. Analysis of MRI brain images¶

Optimized machine learning was able to distinguish very well between ADHD and controls and between persistent and remitting ADHD in young adults using MRI neuroimaging. Relevant features of ADHD compared to controls were identified as:⁹³

• Nodal efficiency in the right inferior frontal gyrus
• Functional connectivity between brain regions right medial frontal and inferior parietal
• Volume of the right amygdala

The remission of inattention and hyperactivity/impulsivity correlated with

• Higher nodal efficiency right medial frontal
• Reduced functional connectivity between right medial frontal and inferior parietal brain areas

5.8. Analysis of urine values¶

A sufficiently accurate analysis of urine values to diagnose ADHD has not yet been reported. One study also found no connection between urine values and ADHD in dogs.⁹⁴

A meta-analysis of k = 210 studies found statistically significant correlations for the following urine values:⁹⁵

• Noradrenaline
  • ADHD
  • ADHD drug reaction
  • Severity of ADHD symptoms
• MPHG (3-methoxy-4-hydroxyphenylethylene glycol)
  • ADHD
  • ADHD drug reaction
  • Severity of ADHD symptoms
• MAO (monoamine oxidase)
  • ADHD
  • ADHD drug reaction
  • Severity of ADHD symptoms
• Zinc
  • ADHD
  • Severity of ADHD symptoms
• Cortisol
  • ADHD
  • ADHD drug reaction
  • Severity of ADHD symptoms
• PEA
  • ADHD drug reaction
• Ferritin
  • Severity of ADHD symptoms

Trends (not statistically significant) were shown for⁹⁵

• Neuropeptide Y
• Manganese
• DHEA (dehydroepiandrosterone)

5.9. Analysis of eye values¶

5.9.1. Electroretinography (ERG)¶

One study found evidence of gender-specific biomarkers in electroretinography in ADHD.⁹⁶

5.9.2. Pupillometry¶

Pupil diameters at rest are a biomarker for tonic noradrenal infiring. Pupil changes during tasks and in response to stimuli are a biomarker for phasic noradrenal infiring. Find out more at* Tonic and phasic noradrenaline* in the article Noradrenaline in the chapter Neurological aspects of ADHD.
One study achieved a sensitivity of 77.3% and a specificity of 75.3% in the diagnosis of ADHD compared to a healthy control group using a machine learning method from pupil measurements.⁹⁷

Several studies found no evidence for the sometimes reported correlation between pupil diameter and fluid intelligence.⁹⁸⁹⁹

5.9.3. Eye tracker: duration of target fixation (?), duration of gaze on irrelevant areas¶

The study of distractibility by recording and analyzing eye movements during a task-irrelevant distraction in children and adolescents aged 6 to 17 years showed a significant (negative) correlation between a (reduced) duration of target fixation and (increased) attention problems reported by parents (p < 0.001).¹⁰⁰ However, 5-month-old children who were later diagnosed with ADHD showed a positive correlation of prolonged gaze fixation during an attention task (21.2 seconds) compared to the 5-month-old children who later did not show ADHD (12.7 seconds). As only the group values were so clear, diagnostic use was not possible.⁵⁸

Two studies report on prolonged viewing times of irrelevant areas during a continuous performance task, which could have diagnostic value as additional information ¹⁰¹¹⁰²

5.10. Activity measurement / actigraphy¶

A very exciting study reports a highly significant measurement of inattention and hyperactivity in a school class (n = 39) using actigraphs (accelerometers) over 3 weeks.¹⁰³
The measurement of activity synchrony found a negative correlation between synchrony of movements and inattention, but not hyperactivity (in the teacher’s report).
The measurement of activity volume correlated with hyperactivity (in the parent report), but not with inattention.

Other small studies dealt with the diagnosis of hyperactivity using a smartwatch with a special app¹⁰⁴, actigraph measurements for ADHD diagnostics¹⁰⁵¹⁰⁶ , differential diagnostics for Bipolar Disorder¹⁰⁷ and the detection of aggression in ADHD¹⁰⁸

The measurement of pixel changes in a 4-minute video recording to detect ADHD compared to non-affected individuals showed an accuracy of 90.24%, a sensitivity of 88.85%, a specificity of 91.75% and an area under the curve of 93.87%.¹⁰⁹

5.11. Inflammatory biomarkers¶

The inflammatory markers found for ADHD were evenly distributed among the three ADHD presentation forms¹¹⁰
Insofar as the detection accuracy was indicated, the AUC, sensitivity and specificity show that none of the biomarkers determined achieves the quality required for diagnosis. It remains to be seen whether a combination of several values could achieve adequate rates

5.11.1. Platelet-lymphocyte ratio (PLR)¶

The platelet/lymphocyte ratio (PLR) is a biomarker for inflammation.¹¹¹¹¹²

ADHD shows an increased PLR ^{113 114 111 115 116117}

AUC	Cut-off	Sensitivity	Specificity
79 %118	93.18
68 %119	115.7	82 %	52 %
60 %120	130.55	36 %	72 %

ADHD medication, age and gender did not influence PLR. The measure of PLR showed no correlation with the severity of ADHD symptoms.¹¹¹
A study of n = 1,455 children, 91 of whom had ADHD, found an increased platelet-to-lymphocyte ratio in ADHD, particularly in males, children of mothers aged 20-29 years and children with asthma.¹²¹

5.11.2. Neutrophil to lymphocyte ratio (NLR)¶

The neutrophil-lymphocyte ratio is a biomarker for inflammation.¹¹¹¹²⁰

The majority of studies (k = 9) found an increased NLR in ADHD.^{113114 111 115 122 123 116 124 117} and also in ASD, where it correlated with social interaction problems¹²⁴.

AUC	Cut-off	Sensitivity	Specificity
78 %118	1.3
78 %119	3.07	66 %	84 %
66 % only for boys120	1.62	54 %	70 %

Lymphocytes were significantly reduced in ADHD.¹²³¹¹⁶ Atomoxetine increased the lymphocyte count.¹¹⁶
Neutrophils were elevated in ADHD.^116124117
ADHD medication, age and gender did not affect NLR. The measure of NLR showed no correlation with the severity of ADHD symptoms.¹¹¹

Two studies found evidence of reduced NLR in ADHD:

• reduced NLR in ADHD and ASD¹²⁵
• decreased NLR in ADHD and the amount of lymphocytes correlated positively with the severity of ADHD symptoms¹²²

5.11.3. Monocyte/lymphocyte ratio (MLR)¶

The monocyte/lymphocyte ratio (MLR) is a biomarker for inflammation and for ADHD. An increased MLR was found in ADHD:
AUC 57 % with a cut-off of 0.23, sensitivity 45 %, specificity 66 %.¹²⁰
AUC 75 % with a cut-off of 0.13.¹¹⁸

The lymphocyte count in ADHD was significantly reduced.¹²⁰
The monocyte count in ADHD was significantly increased.¹²⁶

A meta-analysis found no change in MLR in ADHD¹¹⁴, as did a more recent study¹¹⁷.

5.11.4. Mean platelet volume (MPV)¶

The mean platelet volume (MPV) is a biomarker for inflammation.¹²⁰
The MPV was significantly increased in children with ADHD (n = 152)¹²⁰¹¹⁷ , as well as in children and adolescents with ADHD (n = 552)¹²⁷.

AUC	Cut-off	Sensitivity	Specificity
68 %119	11.1	38 %	98 %
66.2 % only for boys128.	9.75	71.6 %	55.9 %
60 % for children120	9.45	56 %	59 %

5.11.5. Platelet distribution width (PDW)¶

PDW was elevated in ADHD.¹¹⁵
PDW correlated with hyperactivity symptoms.¹²⁹

5.11.6. Plateletcrit (PCT)¶

Plateletcrit (PCT, platelet hematocrit, platerethritis) is the percentage of platelet mass in the blood volume. Calculated from platelet count (PLT) and mean platelet volume (MPV). Provides information about the total platelet mass and is an important indicator of disorders of primary hemostasis. PVT can help to detect quantitative platelet abnormalities such as thrombocytopenia (low count) or thrombocytosis (high count) even if the platelet count appears normal.
The PCT correlated moderately negatively with ADHD symptoms.¹²⁹

5.11.7. Leukocyte count (MBC)¶

The white blood cell (WBC) count was elevated in ADHD.¹¹⁵

5.11.8. Basophil count¶

An increased basophil count correlated with ADHD, but only in boys. The AUC reached 65.6 % at a cut-off of 0.04, sensitivity 70.4 %, specificity 54.2 %.¹²⁸.

5.11.9. Eosinophil count¶

The serum eosinophil count was elevated in children with ADHD¹³⁰, and drastically so.⁷
Eosinophils are white blood cells that are involved in the body’s immune response.

5.11.10. Anti-myelin proteins¶

One study found significantly increased levels of in drug-naïve children with ADHD aged 8 to 14 years:¹³¹

• MAG (myelin-associated glycoprotein)
• CDNF (cerebral dopamine neurotrophin)
• hs-CRP (highly sensitive C-reactive protein)
• Reelin
• Cerebellin-1

Unchanged were:

• Anti-MBP (anti-myelin basic protein)
• Anti-MOG Anti-myelin oligodendrocyte glycoprotein)

5.11.11. Neutrophil-to-high-density lipoprotein ratio¶

Medication-naive children with ADHD showed an increased neutrophil-to-high-density lipoprotein ratio. MPH improved this.¹³²

5.11.12. Lymphocyte-to-high-density lipoprotein ratio¶

Medication-naïve children with ADHD showed an increased lymphocyte-to-high-density lipoprotein ratio. MPH improved this, except in ADHD-C, where it increased.¹³²

5.11.13. Monocyte-to-high-density lipoprotein ratio¶

Medication-naive children with ADHD showed an increased monocyte-to-high-density lipoprotein ratio. MPH improved this.¹³²

5.11.14. Platelet-to-high-density lipoprotein ratio¶

Medication-naïve children with ADHD showed an increased platelet-to-high-density lipoprotein ratio. MPH improved this.¹³²

5.11.15. CRP¶

CRP is a general inflammatory biomarker.
CRP was elevated in children with ADHD or ASD, but did not correlate with symptom severity.¹²⁵ CRP correlated with procalcitonin.

5.11.16. Procalcitonin¶

Procalcitonin (PCT) is an inflammatory biomarker released by parafollicular thyroid cells
Procalcitonin was elevated in children with ADHD or ASD, but did not correlate with symptom severity.¹²⁵ Procalcitonin correlated with CRP.

5.11.17. Systemic immune inflammation index¶

Medication-naïve children with ADHD showed an increased systemic immune inflammation index (SII).¹¹⁶ MPH improved this.¹³²
Another study found no evidence of changes in the systemic immune inflammatory index (SII) in ADHD, but evidence of increased SII in hyperactivity¹³³

The pan-immune inflammation value (PIV) was also found to be elevated in ADHD¹¹⁶

5.12. Intrinsically disordered regions of active proteins (IDR)¶

Intrinsically disordered regions are biologically active protein regions that have high conformational variability but no stable three-dimensional structures. Intrinsically disordered proteins (IDP), which contain IDRs of different lengths or are completely disordered, are of central importance in many biological processes, such as¹³⁴

• Signal paths
• Transcription
• Translation
• Cell cycle

The precise control of the numerous intrinsically disordered proteins (IDPs) in cells is important for proper signal transmission. Mutations or changes in IDPs can lead to various diseases,¹³⁴

The proportion of IDPs in ADHD (73.4 %) and ASD (75.4 %) was increased (normal value 66.2 %), unlike in anxiety (significantly reduced here) or bipolar, depression, schizophrenia or mental retardation (little different here).¹³⁴ There were also significantly more proteins with disordered binding sites. In addition, ADHD (and even more so ASD) showed a significantly increased proportion of LLPS proteins (liquid-phase-separated proteins).

IDPs for ADHD affect¹³⁴

• the activities of various ion transmembrane transport proteins
• the regulation of the membrane potential
• the regulation of the ionic transmembrane transporter
• the regulation of synaptic signal transmission

IDP for ASS concern¹³⁴

• molecular functions, especially those closely related to the regulation of gene expression, such as
  • Chromatin binding
  • DNA binding
  • Binding of DNA-binding transcription factors
• biological processes
  • Learning and memory
  • chemical synaptic regulation
  • Behavior
  • Head development
  • Cognition

The proteins prominently named by Zhang et al. with a high involvement in mental disorders have all already attracted attention in connection with ADHD.

• TP53 is degraded by the protein encoded by the ADHD candidate gene TAF1
• CREB1 is encoded by a candidate ADHD gene
• CREBBP is known to be a monogenetic cause of hyperactivity
• KRAS: The protein encoded by the ADHD candidate gene RASSF4 can act as a KRAS protein

5.13. Metabolite profiles¶

Several studies have looked for metabolites typical of ADHD.
In the blood serum, statistically significant values were found for 156 metabolites examined:¹³⁵

• Increased
  • Cholic acid
  • Homoveratric acid
• Reduced
  • Nicotinic acid
  • Inosine

In the blood plasma of children with ADHD, a study found¹³⁶

• Increased
  • Guanosine (but did not correlate with ADHD symptoms)
  • O-phosphoethanolamine
  • Phenylleucine (but did not correlate with ADHD symptoms)
  • Hypoxanthine (but did not correlate with ADHD symptoms)
  • 4-aminohippuric acid
  • 5-Hydroxylysine
  • L-cystine
• Reduced
  • Gentisic acid
  • Tryptophyl phenylalanine

Urine and fecal samples from a cohort of twins with and without ADHD found¹³⁷

• in the urine of men
  • an increased excretion of
    • Hippurate (correlated positively with ADHD and negatively with IQ)
• in the stool of people with ADHD
  • an increased excretion of
    • Stearoyl linoleoyl glycerol
    • 3,7-Dimethylurate
    • FAD (flavin adenine dinucleotide)
  • a reduced excretion of
    • Glycerol-3-phosphate
    • Thymine
    • 2(1H)-quinolinone
    • Aspartate
    • Xanthine
    • Hypoxanthine
    • Orotate
• many of these intestinal metabolites had a stronger genetic influence than environmental factors

5.14. Language analysis¶

5.14.1. Analysis of voice recordings¶

An analysis of voice recordings was able to identify ADHD in young women with an accuracy of 87% (AUC). The accuracy was lower in older people. However, detection was independent of existing comorbidities.¹³⁸

5.14.2. Stylometric features¶

Stylometry (style statistics) is the quantitative evaluation of linguistic and textual stylistics.
Adolescents with ADHD produced narratives that were shorter, less lexically diverse and less cohesive. Stylometric analysis using an SVM classifier distinguished between the ADHD group and the control group with an accuracy of up to 100%. Clear linguistic markers were found that may reflect difficulties in emotion regulation.¹³⁹