Child First Project

Effective Decision-Making Support for Intervention of Child Abuse and Neglect: Using Information-Communication-Technology

Other language versions (such as, Arabic, Chinese, French, Greek, Portuguese, Russian, Spanish) are available.

Please check out "Volume 23, Number 3, December 2014" in the following URL.

http://www.ispcan.org/?page=Link

 

Background

 

         The number of child abuse cases handled by Japan's Child Guidance Centres in fiscal 2013 stood at 73,765, with the social costs of child abuse in Japan calculated at around $16 billion (USD). However, the child welfare budget in Japan and the number of professionals were limited in relation to developed countries such as Canada and the U.S. (Wada & Igarashi, 2014). Currently, Child Guidance Centres in Japan are introducing multi-disciplinary team (MDT) approaches which collaborate information and practices regarding children’s welfare, medical professionals/institutions, and forensic professionals and interview skills. Furthermore, in Japan, no integrated database for child abuse and neglect exists among the Child Guidance Centres, local governments, and the police. This has lead to problems with how Japanese child abuse and neglect professionals deal effectively with poly-victimization (Slowikowski, Finkelhor, & Hamby, 2011) and how they apply evidence-informed practice (Shlonsky & Mildon, 2014). To solve these problems, our teams along with other colleagues in the child abuse field are not only conducting effective evaluations on decision-making and cost-analysis, but are also in the process of building a brand new practice-research design for real-time analysis and feedback. This is to create a prospective recommendation for decision-making using modern technological approach.


         

Recommendation
system for Child safety


        This recommendation system plans to use the latest evidence available, with the goal for professionals to avoid subjective decision-making. It works similarly to the sabermetrics system in baseball, as seen in the movie Moneyball (2011) starring Brad Pitt. This system is expected to drive professionals’ decision-making speed by providing real-time recommendations. This will depends on each case, along with their culture, geographical area, and ethnicities. This concept is sustained through progressive development. Continuous data analysis and input from multiple professional practices (medical, forensic, etc.) is streamlined into a central database in order to keep the latest evidence and provide the most up to date recommendations. 

Practice & Research
circle for the best practice in future.


     It is important for MDTs to share similar backgrounds and knowledge when they discuss their child abuse cases in meetings (Bahrami et al., 2010; Kim & Mauborgne, 2003). Therefore, this project aims to create an investigation application so all practitioners can use the large amount of database evidence at appropriate times in their field with real-time analysis. In order to recommend the best information to effectively support their decision-making related to child safety, each result of practitioners’ decision-making should be added and updated to the database automatically. Such a database can be useful for practitioners because a practitioners’ decision-making history adds his/her experience to the database. Also with larger sample sizes, the database keeps increasing reliability of the recommendations. 

1. Creating an Application

        If practitioners try interviewing a child multiple times, the child may possibly recant their disclosure because interviewers may provide suggestibility, and can cause additional stress to the child (Ceci, HuffMan, Smith, & Loftus, 1994; London, Bruck, Ceci, & Shuman, 2005). So it has become apparent that reducing the amount of interviews is important. To avoid stressful situations for the children, this application will use an age-based standard interview protocol for all practitioners. These interviews refer to forensic interview skills and knowledge comparable to the Child First Protocol (National child protection training center, 2014), and National Institute of Child Health and Human Development (NICHD) protocol (Lamb & Garretson, 2003). This application will not only have interview protocol, but also be able to reference photo evidence of bruises and injuries that resulted from child abuse. Including photographs for reference will help to strengthen forensic evidence regarding each case. Furthermore, with the functionality to add a local risk assessment tool and reporting function to the application, users will be able to find the closest child protection centre using GPS API. We have already created a draft version of the interview guide and a photo evidence wizard. A current demonstration video on YouTube is also available.[1]&[2] Our draft for the application received the People’s Choice Award and the MGS Design Award at the 2014 Hacking Health Hackathon in Vancouver.

 

[1]Hacking Health Hackathon Vancouver 2014 Live Pitch (2 mins)  (https://www.youtube.com/watch?v=5fSuHt17Ln4)

 

 

 

2. Building Cloud Databases

        In order for this application to be successful the information will need to be stored in a secure database. This application and database should have the latest encrypted security system and accessibility to proper authorities. Using a cloud database, all practitioners would easily be able to access and share the abuse case information. In the U.K., there are current discussions that would allow emergency medical doctors to have access to their child abuse database (from the child protective services) starting in 2015 (Hawkes, 2012).

        Additionally, if socio-demographic information and case-logs (not containing personal contact information) could be standardized and preserved on the cloud database, researchers could easily conduct text-mining and chronological analysis. They could potentially create models of successful and/or failed MDT collaborations which would guide future approaches towards each child abuse case. Practitioners could utilize similar cases, predict the estimated duration of the case, and refer to the decision-making that past practitioners had and their results. Along with socio-demographic information and case-logs, photo evidence could be used to compare and predict recurrent rates and fatality rates as prospective research. Accessibility will be vital for practitioners and researchers in order to continually update the cloud database for real-time analysis.

3. Improving Data-Analysis Algorithms

        The main purpose of this project is to provide an objective recommendation using the latest and most relevant research evidence that is provided to support all practitioners’ decision-making. An improved data-analysis algorithm would then be necessary to facilitate these means.

        If there is a large amount of samples in the database, an algorithm of logistic regression analysis can refer to previous research designs about recurrent rates and fatality rates (Sledjeski, Dierker, Brigham, & Breslin, 2008).

If there is a limited sample size, we plan to implement the algorithm of Bayesian network analysis to predict children’s safety probability in each case (Proeve, 2009). Bayesian network analysis uses an inductive approach that starts from a small amount of data, and as the sample size increases it will eventually be used in conjunction with deductive logistic regression analysis. It can be used to initially predict the probability of improving parenting attitudes and find more effective clinical procedures pertaining to aspects such as cost effectiveness, and the shortest duration for investigation procedure for MDTs.

        This Bayesian network’s derived predictions are similar to how recommendations on the website amazon.com are created. There, the more you buy products, the more reliable the recommendation becomes. This system which we also hope to utilize is called collaborative filtering.

4. Updated Clinical Practice with Recommendation Support

 

       All final decisions should not be made by computer programs but the practitioners themselves. However, so as to keep enhancing the reliability of the recommendation system, we need to compare the provided recommendation with the actual decision, and continuously record all results into the database. When deviations from a typical decision-making process are made, it becomes notably important to archive this data to see if an innovative method should be applied instead. This also applies to what methods practitioners should be cautioned for.

        Recommendations could develop various perspectives of supervision. This database could potentially be accessed by practitioners of institutions in rural areas, to which they can refer the decision making that practitioners did in urban areas. To put it another way, the recommendation might be able to provide a function of distance learning.

        In addition to providing information from and to their peers, the database would be available for practitioners to reflect their own history of decision-making. They can view their profile and learn about their positive and negative tendencies when they make their decisions. Education and training for practitioners are long term investments, and our project can act as a kind of education feedback system for them. This type of feedback could lead to more cost-efficient performance. In near future, we plan the recommendation run on wearble devices and provide appropriate sentences based on clinical psychology knowledge to tell children. 

Conclusion


        Our team is attempting to create this application and database with Department of Engineering at the University of British Columbia. On top of creating the proposed algorithms, we need to ensure the best security options are implemented. If we could use a secure database in universities, researchers can cooperate with local practitioners to evaluate their programs as prospective studies. This is how our project will begin and eventually the collaborations between various practices, research, and education will appreciate into the necessary foundation for our project. The method of Bayesian network analysis for probabilities and recommendations will be an innovative thought process in this field, as only one child abuse paper has been published using the Bayesian network. Using latest technologies can help drive this progress and help develop our own technology so our project can create evidence and recommendations with high validity as soon as possible.

        In Japan, the field of child abuse falls years behind in comparison to other countries. While our project’s initial goal is to create a central database and standardize methods in Japan, it is our goal that our application can evolve to accommodate other countries and help improve and stabilize their decision-making processes. As seen in the previous figures, harmonious MDT action and support is the key to the success of our entire project. After private discussions with various child abuse and neglect professionals, we have found encouraging support for this project. 

 

Copyright (c) 2014 "Child First Project" Kota Takaoka All Right Researved. 

Illustration copyright (c) 2014 @childfirsttan & Kanako Ogura All Right Researved.

Picture images are quoted from ABFO#2 websight. WIX free images.
Wearble device images are quoted from Google grass.

Kota  
Takaoka

Hello!

 

I'm Kota Takaoka. 
 

Please feel free to leave your message.