The term Attention Deficit Hyperactivity Disorder (ADHD) is used by the American Psychiatric Association in the Diagnostic and Statistical Manual of Mental Disorders [1
] to describe a persistent pattern of inattention and/or hyperactivity-impulsivity that is inconsistent with developmental level, and that impacts negatively and directly on social and academic/occupational activities. We are thus situated in the last interpretation of an old problem that, although highly topical, has a long history of queries behind it.
The problems related to ADHD not only come from outside the scientific community, but have also spread within itself: multiple doubts in reference to etiology, prevalence, evaluation instruments and procedures, etc. The key to all these problems lies in the lack of agreement when considering diagnosis. Thus, in the absence of any biochemical, structural or genetic condition that unequivocally determines the existence of ADHD, the diagnosis is clinical, that is, based on the professional expertise of the doctor, and it is determined by the observation and information provided by parents and teachers [2
]. This is highly subjective and leads to disparate results, largely due to the lack of agreement in the assessment procedures and evaluation instruments [3
]. In addition, as it has been stated by the Spanish Association of Neuropsychiatry, “the main problem is that the clinical criteria used to diagnose this disorder are too vague. The diagnostic manual used, DSM-V, includes very broad and subjective criteria. How can be determine if a child is more or less prone to a higher degree of movement?” [4
Consequently, the inaccuracy of the diagnosis of ADHD, based on subjective criteria [5
], together with the fact that hyperactivity is one of the main symptoms of this disorder [6
], means that, for more than a decade, several studies have been carried out to record objective measures of movement in the subjects. However, these studies have a number of limitations. First of all, the use of accelerometers (actigraphy and inertial measurement units), that is, that some device has to be placed in the body of the subjects, limiting their ecological validity and, secondly, the use of infrared devices that only analyse some parts of the body [9
The present paper tries to fill the aforementioned shortcomings by using Microsoft Kinect V.2. This device is able to completely track and record 25 joints of six human bodies through an infrared system, without placing any type of sensor in the body of the subjects [14
]. The commercialized version of this device allows developers to make their own programs by using gestures and body movements in a wide variety of applications (Ding and Chang [20
]) besides, some emerging researches on the use of this device (that enable us to count limb movements in subjects with ADHD) conclude that it is a good device to measure movements [21
]. Furthermore, since this is a device of small dimensions, it can be introduced into the natural environment of the subjects.
For all the above, and trying to contribute to broaden the knowledge about the behaviour of subjects with ADHD, a computer program has been developed to objectively record the amount of movement of the subjects. Subsequently, study techniques workshops have been designed as an incentive used for attendance, both for subjects with a firm diagnosis of ADHD and for control subjects.
The main objective of this study is to introduce a multidisciplinary tool developed for objective analysis of movement in subjects with ADHD. A computer program was developed to record the amount of movement of subjects. The validity of this software as a tool to support the diagnosis of ADHD has been carried out by comparing the amount of movement of the group diagnosed with ADHD and the control group, and it has also been compared with the movement registered by observers.
The identification of movement, posture, or a gesture made by a human body in real time is a difficult challenge because it has been found that the human body can perform a great deal of movement and it can have different size etc. [26
]. Notwithstanding the above, after having carried out a series of tests with different technologies, the analysis of various technological devices, and the confirmation that Kinect has been previously used successfully in a wide range of research fields [27
], it has been concluded that the Kinect V.2 is a suitable device to recognize and capture the movement of subjects in a teaching/learning situation, and thus achieve the objectives of this research.
Therefore, after having analysed existing technology, the computer application ADHD Movements was developed to detect the number of visible skeletons and, from each of them, analyse and draw on the screen each of the joints. From each joint that makes up all the skeletons, the application estimates the distance it covers in the 3D space during each minute. So, for each joint, the distance between the different 3D points of the trajectory of the joint in the movement has been calculated for each minute of the session.
Once the computer application was developed and verified, the movement of the subjects was captured.
Results show that there were significant differences in the amount of objective movement between a clinical group of subjects with ADHD and a control group, obtaining a higher average of movement in the experimental group in all the analysed joints (except in the left ankle joint). In addition, the differences found between the two groups are statistically significant in 14 of the 17 joints analysed.
In the left ankle joint, the control group obtained a higher movement average than the experimental group, although there was no statistically significant difference. This may be due to the dominance or preference for the use of the right or left leg of the subjects that have been part of the research groups.
From the revision of the literature it is concluded that this is the first study that counts the amount of movement in different joints with Kinect in order to find differences between a group of subjects with a firm diagnosis of ADHD and a control group. However, given the importance of hyperactivity in ADHD [7
], objective measures of movement have been studied for more than a decade, although motion capture has almost always been done with accelerometers (actigraphy and inertial measurement units), with infrared systems placing some device in the body of the subjects or analysing only a part of the body [6
In all the studies reviewed, children with ADHD showed more body movement than those without ADHD [6
] so the results of the study coincide. As for the objective movement difference estimated in the group of subjects with ADHD with and without medication intake, the results show that there are statistically significant differences for the 17 joints analysed, being in all of them the average of movement higher in the group without taking medication. The results we have obtained are similar to those of previous studies that indicate that medicated ADHD children show significantly less motor activity than non-medicated ADHD children [41
Along the lines of a recent study in which the effect of methylphenidate was studied in children with ADHD [44
], and in which head movements were not found to be significantly different between the groups (medicated ADHD, non-medicated ADHD and control children), in the study, in the head
joint, although the difference found between both groups is statistically significant, it does not obtain the minimum levels required by Cohen to be considered relevant [25
The non-specificity of some diagnostic criteria defined by the DSM-V and the WHO for ADHD, such as squirm in their seat
, means that although applications have been developed specifically to estimate the quantity of movement [45
], the estimation of this depends on what the observer associates with the gesture. For this reason, both the movement to leave the seat
, as well as the one previously mentioned, in the study, were registered by different observers.
Limitations, Strengths, and Future Directions
It is necessary to mention that this work is not exempt from limitations that must be considered when interpreting the results and their implications, as well as in the preparation of future studies in order to solve them and increase the understanding of the findings.
Firstly, the number of participants in the experimental group (32) may restrict the ability to detect significant differences between the groups. This limitation must be taken into account nevertheless minimized if we consider that other studies have presented a lower sample size [46
]. In addition, the experimental group consisted of 24 boys and 8 girls, that is, 75% of the sample of participants with a firm diagnosis of ADHD were male. This data is in accordance with current epidemiological studies that indicate that the prevalence rate is greater in male [48
]. However, it limits the extrapolation of the results analysed according to the sex of the participants. For this reason, in order to check whether the results are scalable, it would be interesting for future research to use large samples of subjects.
Besides, related to the sample size is the fact that experimental group consisted of 24 subjects with a diagnosis of ADHD with a combined presentation and eight with a predominantly inattentive presentation, according to the DSM-V criteria. In this study we have only analysed hyperactivity as a core feature of ADHD [7
], together with the conditions of compliance with the diagnostic criteria in order to classify a predominantly inattentive presentation in ADHD that excludes those related to hyperactivity. Therefore, it would be interesting that in the future the objective movement tests carried out in this study could be carried out also in ADHD subjects with a predominant hyperactive/impulsive presentation.
Secondly, in this study the Kinect V.2 sensor was used to record the movement of subjects due to its simplicity and effectiveness [20
]. In addition, different studies carried out with the Kinect device in several fields show that, although this is not as accurate as some traditional measurement technologies in research laboratories, it provides a good quality relationship for motion tracking systems with respect to the length of the body segments, joint angles and the displacement of the joints in the different body gestures [23
]. All the above is not an obstacle to emphasize that the device has a series of limitations that must be considered. Thus, the Kinect device shows excellent results in wide movements such as sitting or standing up, but it shows poor precision in fine or small movements such as closing a hand [49
]. This fact has led us to dismiss the results of the hand
, hand tip
joints. On the other hand, the Kinect device detects and records the human body better in standing position and in the study the data has been captured in real education/learning situations, where the subjects are sitting. For this reason, it is possible that in certain positions of the subjects (for example, crossing the legs) occlusions have occurred. In these cases, the skeleton detection software of the device interpolates the positions of the undetected joints, being able to produce slight deviations between the real position of a joint and the interpolated position.
The results indicate that for the two registered movements, the highest average corresponds to the experimental group with respect to the control group, although only in the movement squirm in seat the difference in averages is statistically significant.
In the same way, when comparing the movements registered by the observers between the experimental group with and without the medication, the highest average corresponds to the group without medication. Only in the movement squirm in seat the difference of averages is significant, with a high magnitude (d = 0.80).
The absence of significant differences in the movement leave their seat when comparing the experimental group with the control group, may be due to both the “novelty” of the teaching/learning situation (it was the first time they were in that classroom and with those speakers) or to the effect of the medication. However, the results of the comparison of the group with and without medication cannot be due to this reason, since the assistants already knew the speakers and the classroom of the study techniques workshop.
In any case, in the absence of previous studies in which a register of this type of movements associated with the diagnostic criteria of DSM-V and WHO for ADHD in a teaching/learning situation, we cannot compare results.
The above results lead to a series of conclusions:
The software developed (ADHD Movements) for the Microsoft Kinect V.2 device is valid to capture the movement of 17 joints of up to 6 subjects in a teaching/learning situation.
Students with ADHD present more movement and squirm more in their seat, than students without ADHD.
Students with a firm diagnosis of ADHD without the prescribed medication present more movement and squirm more in their seat than ADHD students with the prescribed medication.
ADHD students with and without taking their prescribed medication present a similar amount of movement in the head joint.
Girls with ADHD present more movement than boys with ADHD.