A NEUROCOGNITIVE PROTOCOL SYSTEM TO SUPPORT

HEALTH AND CARE OF ABUSED CHILDREN

Carlo Emmanoel T. de Oliveira, Carla Verônica M. Marques

Programa de Pós Graduação em Informática, Universidade Federal do Rio de Janeiro

Av. Athos da Silveira Ramos, 274, Rio de Janeiro, Brasil

Jorge Neval Moll Neto

Unidade de Neurociência Cognitiva e Comportamental, Rede LABS-D'Or Hospitais, Rio de Janeiro, Brasil

Keywords: Neuropsychology, neurobiology, abused children, assessment, datamining, decision support systems, web

development.

Abstract: Abused children is highly endangered of developing critical cognitive dysfunctions. Clinical observation

has encountered many related cases of abuse and poor learning performance. Authorities unaware of these

conditions may take longer to act, detrimentally to the child welfare. This work provides a wide coverage of

medical protocols for every area concerned with endangered children procedures. These protocols were

researched with the collaboration of specialist in each area to achieve the most detailed and conspicuous

information of children status. These protocols are proposed as a Web system available to all concerning

professionals and authorities to input and access the relevant information. This data can be processed and

analyzed to provide decision support and handling indications derived from statistical and heuristic

treatment of the whole information.

1 INTRODUCTION

The violence in the childhood is highly co-related with

serious behavioral, cognitive or emotional damages that

are immediately noticed on school learning, on

language expression and on the relationship life. The

abuse and negligence commit the social cognition

development, that means to say that the semantic and

pragmatic dimensions of the language become very

impaired, what leads to the delay in the vocabulary

acquisition and development and in grammatical

structures of the oral and writing language.

These results found in the neuropsychological tests,

show qualitative alterations, mainly in tasks that

involve the activity of the frontal lobe. On first

analysis, poor performance was observed in

sheltered children, in levels, co-varying the type,

intensity, duration of the abuse and/or negligence,

age group in which elapsed the abuse situation and

presence of aggravating environmental factors

(shelter type or street experience) or the opposite,

opportunities of compensatory interpersonal entails.

Qualitative differences were identified on the

cognitive development, especially in the attention,

formal learning, memory, language, abstract

reasoning and executive functions, without deep

lowering of the global cognitive competence (IQ).

In the same way, the behavioral alterations in

sheltered children, identified as risk factor for the

development of psychiatric impairment, appeared to

be related to the abuse conditions, abandonment or

negligence. This symptoms varied a lot depending

on the anxiety, shyness, phobias, panic syndrome,

impulsiveness, low-self-esteem, little or no tolerance

to the frustration, disturbed sleep, night enuresis and

presence of psychogenic motor stereotypies.

Behavior disorders - post-traumatic stress

syndrome, deficit of attention disorder, psychotic

symptoms, obsessive-compulsive disorder and

conduct disturbances, little interpersonal ability,

acceptance anxiety, imitative behavior, and even

dissociatives disorders: hiperactivity, hipervigilance,

threats illusory perception, paranoiac interpretation

of interpersonal relationships, emotional immobility,

incapacity of reacting when challenged or in

127

Emmanoel T. de Oliveira C., Ver

onica M. Marques C. and Neval Moll Neto J. (2008).

A NEUROCOGNITIVE PROTOCOL SYSTEM TO SUPPORT HEALTH AND CARE OF ABUSED CHILDREN.

In Proceedings of the First International Conference on Health Informatics, pages 127-132

 SciTePress

pressure, inferiority and uselessness feeling, suicidal

and/or homicidal ideation, hallucinations, irritation,

despair and self-mutilation.

And finally dysthymia: chronic depressive state,

melancholy, obsessive-compulsive behavior, bad-

humor, low motivation, low self-esteem, emotional

apathy, pessimism, anxiety, chronic fatigues, self-

isolation, alimentary and drug addiction disorder.

We still have academic difficulties in the

learning of school contents in the Portuguese

language and mathematics areas.

Our investigations show that sheltered, abuse or

negligence victims, children have low performance

at school, revealing that the consequences of

violence can persist years after their retreat from the

streets or from the origin family. According to the

collected information, it happens because the child's

past traumatic experiences seem to be added to the

sheltering situation in which they are. Besides to

total absence of specific educational programs for

these children, they make their formal learning

impossible, becoming an apart group inside the

school. The life conditions in the shelters become an

aggravating factor on symptoms of cognitive delay

and school abandon, appearing as the cause of high

stress and propitiating the continuity of the privation

conditions and abuse, that end up resulting in the

escape of children from shelters and in the

consequent school escape. As well as the school

reaffirms these differences becoming the official

organ that decrees the failure and the social

exclusion.

2 OBJECTIVE –SUPPORTING

ENDANGERED CHILDREN

TUIA is a computational program constituted of a

base of organized data to recover and to co-relate

information associated to the abuse, abandonment

and negligence experiences against the child or

adolescent. It was projected to facilitate the interaction

among specialists of different areas interested in child

abuse, and also encouraging researchers to collect

information from multiple services. The compilation of

these intersubject information allows to cross check

information from the various data sources, reinforcing

more subtle indications. Cross checking medical and

neurocognitive data can exemplify the case. Data

measuring child abuse experience from medical

sources can be co-related the neurobiological

performance (De Bellis MD, 2005),

neuropsychological (Beers SR, De Bellis, 2002),

educational and the child's physics. The combination

between the two modules aims to find the sequels

indexes, esteeming the relative direct consequences to

the abuse, abandonment and negligence indexes.

The program enhances the handling endangered

children, allowing the concomitant examination of

multiple aspects. Traversal of the whole information

database can unravel hidden indications, which are

not directly accessible, but can be only inferred

through the intelligent comparison of the parts. An

information system integrated in a national ambit,

should be projected to be used jointly with the

specialists and authorities, promoting an appropriate

sequence of the institutional actions for prevention,

identification, evaluation and attendance of this

population (Stemberg,K.J.,2004).

3 REVIEW – THEORETICALS

There is now abundant evidence that childhood

abuse and neglect can result in permanent changes to

the developing human brain. These changes in brain

structure and function appear to cause psychological

and emotional abnormalities during childhood and

adulthood. Behavioral and psychological problems

include impulsive and instrumental aggression,

learning disabilities, mood disorders, post-traumatic

stress disorder and antisocial personality disorder,

among others.

Modern neuroimaging techniques, among which

magnetic resonance imaging (MRI) methods stand

out as the most promising ones. MRI not only is

non-invasive (no ionizing radiation or other

biological effects have been shown), but it provides

powerful ways to directly address brain structure and

function in detail both in transversal and in

longitudical studies.

Voxel-based morphometry, for example, is an

especially interesting technique which allows the

study the structure of virtually the whole brain in a

statistically robust manner. By comparing control

and patient groups paired by demographic variables

(e.g., age, education, physical developmental

measures), statistical maps of grey or white matter

changes can be generated (Good et al., 2001).

Furthermore, continuous variables can be entered as

parameters, allowing for testing not only categorical

differences among groups, but also dimensionally.

Voxel-based-morphometry has been successfully

employed to detect subtle anatomical changes in

neurologic and psychiatric disorders, such as major

depression (increased amygdala volume), antisocial

personality disorder (reduced anterior temporal lobe

HEALTHINF 2008 - International Conference on Health Informatics

128

volume), among others (for a review, see Meyer-

Lindenberg and Zink, 2007). Additionally, a new

MRI technique dubbed diffusion tensor imaging has

been developed during the past 10 years. This

modality allows the determination of microscopic

water molecule flow (Brownian movement) in the

living brain tissue; further developments have used

mathematical models of water diffusibility, and now

enable researchers to trace the 'brain wiring', or the

white matter fibers, based on inferences from

preferential water diffusibility. This technique

allows for direct visualization of developmental

changes in the human brain secondary to genetic and

environmental factors. Results from our lab showed

that neuroplastic changes can readily be detected

using this method (Tovar-Moll et al., 2006).

Finally, functional MRI is another especially

powerful non-invasive technique with demonstrated

robustness in detecting functional reorganization of

the human brain following developmental and

cognitive-emotional factors. Our group has been

using functional MRI now for about 10 years in the

study of complex cognitive-emotional phenomena,

which include moral judgments and moral

sentiments (Moll et al., 2005). Based on lesion

evidence and functional MRI experiments in adults,

we have formulated a now influential model of the

'moral brain'. Functional MRI and available

cognitive models can now provide important

guidelines for studies addressing cognitive-

emotional development in normally developing and

abused or neglected children. This will certainly be a

fruitful line of investigation for our understanding of

the complex cognitive and emotional disturbances in

these children, which can guide the development of

better assessment and treatment schedules aiming to

prevent further brain damage or ameliorate

established symptoms. Finally, these imaging

techniques can be employed together with genetic

studies (e.g., gene polymorphisms, endophenotypes,

etc), providing an unique opportunity to explore not

only structural-behavioral or genetic-behavioral

interactions, but direct genetic-structural effects (ex.,

Meyer-Lindenberg and Zink, 2007).

4 ORGANIZATION – A

COLLECTION OF

PROTOCOLS

Thirty-two children were appraised in the age group

between 7 to 12 years, residents of Ayrton Sena

shelter, of the city hall of Rio de Janeiro, that

possess official (juridical, technical or

administrative) registration of abuse, negligence or

abandonment history and 32 children of control

group who go to the same schools the sheltered

children go and who live with their biological

families in a common home. There is no suspect or

record, even informally, of abuse or negligence, by

relatives and teachers. This group is similar in age,

sex, socio-economic level and school levelling.

The sheltered children had a battery of

investigation instruments different from the non-

sheltered children referring to the life history, family

context data and to the current psych-social

situation. It was necessary to diversify, using an

abuse checklist (Joseph Pitty) and a specific medical

history assessment, in the first group, and an

inventory of refined qualitative analysis of family

relationships associated to a general medical history

assessment, appropriate for the second group. Such

procedure looked for comparing different

instruments, similar in objectives and adjusting them

to measure similar information in importance,

relative to its different contexts. In this phase, socio-

economic questionnaires and an environmental

analysis questionnaire were also applied to evaluate

the shelter, and the conventional pediatric and

neurological exam.

The exam of psychiatric impairments tracking,

the CBCL - (ASEBA) - Child Behavior Checklist

(CBCL), the Parent Report Form, and the Teacher

Report Form are part of the first evaluation stage. In

the positive cases of this screening, some

information were explored based on chosen question

from K-SADS-PL - diagnoses interview for children

and adolescent between the ages of 6 and 18 –

Brazilian version of Schedule for Affective

Disorders Schizophrenia for School Aged –

Children.

Based on these general data, the children with

mental and sensorial deficiency were excluded. The

children with complications at birth, serious

diseases, lesions and wounds in the head,

internments in ITC, history of comatose state,

previous evaluation of IQ with index below 80,

history of treatments with psychotropics, psychiatric

impairment, alcohol or drugs abuse, dependence or

prenatal exhibition to alcohol or substances, were

not excluded, but considered as fundamental part of

the research by dealing with most of the subjects.

After this stage, a pedagogic evaluation

(CESGRANRIO) of Portuguese and mathematical

language was applied jointly with the complete

neuropsychological battery, including the language

areas (Capovilla battery), moral competence (Moral

A NEUROCOGNITIVE PROTOCOL SYSTEM TO SUPPORT HEALTH AND CARE OF ABUSED CHILDREN

129

Judgment Test (MJT) - Georg Lind.) and cognition

using Wechsler Intelligence Scale for Children

(WISC III), Tower of London (TOL), Children's

Color Trail Test (CCTT), Test of Cerebral

Dominance (BTN) and Span Cores (short term

memory).

The language evaluation was applied in three

sessions of specific abilities exams composed by the

vocabulary Test in images, Test of phonological

discrimination, Test of words and pseudo-words

repetition, Dictation, nomination Test, Test of letters

knowledge, Test of fluency of words, Proof of

phonological conscience, Proof of syntactic

conscience, Test of silent reading competence and

by the Test of understanding of written sentences.

The analysis of these preliminary data served as

foundation for the development of a protocol of

structural (morphological) and functional

investigations of the brain using the magnetic

resonance imaging (MRI) techniques.

5 IMPLEMENTATION – A WEB

SYSTEM FOR CHILDREN

CARE

An effective way of caring for endangered children

is tracking closely all developments in theirs lives

and handling readily this information to the

authorities responsible for their care. Many children

continue exposed to critical situations due the lack

of perception of health and education authorities of

their particular problem. This work propose to

mitigate this risk by gathering comprehensive

information on critical aspects of children conditions

and making this information accessible within a

decision support tool.

The application is a Web based system with

lightweight componentization and a flexible protocol

reconfiguration architecture. The main purpose is to

support the creation and maintenance of clinical

protocols that can gather the most conspicuous

information about endangered children. The

protocols are classified into team packages and into

specialist forms inside each team division. Team

packages can be assemble to concentrate the

practices requiring a specific expertise. The

packages contain items covering the whole area of

the team expertise and consist mainly of engineered

information collectors that can be readily processed

into decision support reports. In the use cases shown

below, researchers can devise protocols and

analysing algorithms to provide the state of art

children care environment. Institutional teams can

apply the new protocols and analyse the reports to

allocate children to the proper treatment.

Figure 1: Main Use Cases.

The tool not only supports mainstream handling

of current children data, but also is ready to

incorporate new research to the immediate benefit of

these children. The tool includes an upgrade

mechanism to upload new protocols and decision

support algorithms to handle new information and

requirements to cover current children handling

needs. The system is designed to keep up with the

evolution of people caring knowledge, supporting

protocol modifications while preserving previously

collected information. The domain model was

designed taking up a thee years study, covering the

various involved areas and specialists which

contribution was analyzed to develop a common

denominator. The simplified diagram below shows

the tool model, supporting the creation of new

protocols capable of gathering children data.

Reporting tools can also be attached to the system to

provide decision making graphs about a resulting

query.

Figure 2: Static Domain Diagram.

HEALTHINF 2008 - International Conference on Health Informatics

130

The system can be hosted in more than one place

and the packages and its information can be

componentized through JSON (JSON, 2007) and

REST (Khare & Taylor, 2004) communication

protocols. Information is transacted across sites

using REST requests devised as a DOM (DOM,

2007) like interface. Since information about a

patient can be entered in the various hosts

maintained by the specialist institutes, each host can

request complimentary information to the local

specialist data. The requested information is

transacted through JSON packets in response to the

REST requests.

The application has been given the name TUIA

(Test Unification for Indications of Abuse) and is

built upon Web 2.0 techniques, to enable simple

construction, social interaction, and componentized

structure. The idea is to provide easy integration

with other systems composing the TUIA system in

special the neurocognitive learning acceleration

objects. These objects are neuropedagogical games

designed to reduce the gaps between strong and

weak cognition functions.

The interdiciplinary data is stored in the system

database and is processed by algorithms in the

machine learning engine. This engine provides

inference about the healthcare subjects and classify

information to help medical, judicial and political

decisions.

The current engine installed is a Bayesian

clustering (Binder, 1978) algorithm using the

Orange framework. As a proof of concept, a initial

set of 44 children was submitted to a clustering

analysis. This groups was submitted to a battery of

language, mathematics and attention tests. The

group is originally composed of 29 children under

state custody and a control group of 15 regular

school children. The initial groups were assigned as

code Alpha for the control group and Bravo for

custody children. After a session of interated

clustering, the engine worked out six groups,

including the original Alpha and Bravo. The new

groups were arbitrarily named Charlie, Delta, Echo

and Foxtrot. This test was a preliminary probe to

determine what kind of information can be inferred,

and how it can be reported to consist a decision

support. The data was automatically assembled into

radar graphs by the engine using matplotlib (Hunter,

2007). The primary results are shown in the charts

below:

Figure 3: Dysfunction distribution graph.

The charts show the strength and weakness areas

as crests and dents. This charts can help to decide

which children needs the most urgent handling and

which areas are more affected. The labels represent

Vocalization, Attention, Writing, Language,

Mathematics, Dyslexia, Planning and Orality.

Critically damaged Alpha and Bravo children need

urgent care and should be highly prioritized in a

special learning program, with most areas affected.

Echo and Foxtrot children are less impaired but show

different education needs with different gaps between

weak and strong areas. Language is in need in both

and Foxtrot needs more in mathematics and orality

areas while Echo have traces of dyslexia. Delta is

strong in most areas, while Charlie being good in all

logical areas, is rather poor in literacy in general. As a

testbed, this experiment gives a good support to the

idea of integrating medicals protocols in a computing

system. The children classification can help devising

a whole overall strategy in dealing with each group,

matching the children with its needs.

These reports will be available on line and can be

requested as a query on a collection of children. This

chart is a sample report with bayesian clustering, but

other reports can be devised. The query leaves a

formated file in disk and the filename is passed to a

program that analyses it and leaves in return the

result chart to be displayed. The system uses a

framework similar to another scientific web

application, called Enviair (Mota et al, 2007)

developed by our team for environment control.

A NEUROCOGNITIVE PROTOCOL SYSTEM TO SUPPORT HEALTH AND CARE OF ABUSED CHILDREN

131

6 CONCLUSIONS

Children may undergo unnecessary suffering when

indications of their actual conditions are neglected,

even after all health care protocols have been

already applied to them. This is the most ubiquitous

cases, where the concerning authorities are not

prompted with the proper information, or even if the

information is provided, it is hard to roam across a

large amount of raw data to figure out what are the

proper measures in each case.

This work proposes a qualitative improvement in

the programs of prevention, teaching and, in the

therapeutic actions and assistances. A web system

provides a tool to validate and implement

methodologies of technical and administrative

intervention. The TUIA system can maintain an

extensive collection of integrated data on

endangered children, from a wide range of

concerning areas. The system not only provides for

immediate communication of children condition

alerts among the the concerned parts, but also can

aggregate relevant and validated research results to

the benefit of community. This can be achieved as

the system is a platform to develop and validate new

protocols and analytical procedures. The

experimental protocols can be applied together with

the mainstream procedures and the results compared

with cross checking from existing data. New

protocols and reports can be then incorporated into

mainstream as the results have scientific relevance.

The system is developed using state of art web

technology to provide fast and consistent

development. A Domain-Driven (Evans, 2004)

approach is applied to shorten the turnaround of new

software releases as new requirements come in to

play to extend the system functionality. A

lightweight interprocess communication technology

supports distributed hosting if required. It integrates

with machine learning engines and other advanced

analytical tools to provide support to machine

assisted inferences on collected data.

The challenge is to congregate the two aspects of

the public power role: to create methodologies of

scientific investigation, inside the services and in the

universities, capturing the complexity of the theme

and to incorporate the actions against violence in the

involved professionals' practices and to articulate

those practices, not only in the section of Health, but

also in the Education practices, Social Attendance

and Justice.

In a general way, this work proposes the

development of an "algorithm" composed by

taxonomy of information transformed in analysis

methods. The observed population is passed through

clustering algorithms, leading to the identification

of children's sub-groups, according to patterns in

input profiles. This discovery opens the possibility

of multiple studies accomplishment that will result

in a rich knowledge on abuse and its co-morbidities

that do intersection with the several abuse types and

classes.

REFERENCES

Beers SR, De Bellis MD. Neuropsychological function in

children with maltreatment-related posttraumatic

stress disorder. Am J Psychiatry. 2002

Mar;159(3):483-6.

Binder, D.A. Bayesian cluster analysis. Biometrika 1978

65(1):31-38

De Bellis MD. The psychobiology of neglect.

Child Maltreat. 2005 May;10(2):150-72. Review.

DOM - Document Object Model. Avaiable at

http://www.w3.org/DOM/ (accessed July 15, 2007).

Evans, E. Domain-Driven Design: Tackling Complexity in

the Heart of Software. Addison-Wesley, Boston, 2004.

Good CD, Johnsrude IS, Ashburner J, Henson RN, Friston

KJ, Frackowiak RS. A voxel-based morphometric

study of ageing in 465 normal adult human brains.

Neuroimage. 2001 Jul;14(1 Pt 1):21-36.

Hunter, J. D. Matplotlib: A 2D Graphics

Environment.Computing in Science and Engineering,

vol. 9, no. 3, pp. 90-95, May/Jun, 2007.

JSON - JavaScript Object Notation. Avaiable at

http://www.json.org/ (accessed June 9, 2007).

Khare, R., Taylor, R.N. Extending the Representational

State Transfer (REST) architectural style for

decentralized systems. ICSE 2004. Proceedings. 26th

International Conference on Software Engineering,

2004.

Meyer-Lindenberg A, Zink CF. Imaging Genetics for

Neuropsychiatric Disorders. Child Adolesc Psychiatr

Clin N Am. 2007 Jul;16(3):581-597.

Moll J, Zahn R, de Oliveira-Souza R, Krueger F, Grafman

J. Opinion: the neural basis of human moral

cognition. Nat Rev Neurosci. 2005 Oct;6(10):799-809.

Mota, L.C; Oliveira, C.E.T; Meirelles, M.S.P; Berroir,

J.P.; Herlin, I. The Implementation of a web-based

System for Automatic Classification of land use and

covering changes. IADIS International Conference

Applied Computing: Salamanca, 2007.

Tovar-Moll F, Moll J, de Oliveira-Souza R, Bramati I,

Andreiuolo PA, Lent R. Neuroplasticity in human

callosal dysgenesis: a diffusion tensor imaging study.

Cereb Cortex. 2007 Mar;17(3):531-41.

Sternberg KJ, Knutson JF, Lamb ME, Baradaran LP,

Nolan CM, Flanzer S. The child maltreatment log: a

computer-based program for describing research

samples. Child Maltreat. 2004 Feb;9(1):30-48.

HEALTHINF 2008 - International Conference on Health Informatics

132