Classes

Predictive Risk Tool

Predictive Risk Tool (demo): https://gerardchung.shinyapps.io/prm_demo/

Jordan Case study

A social worker call in to your hotline about Jordan who found to have caning marks on his body.

Jordan, a bright and curious 4-year-old, lives in a small, close-knit community with his single mother, Elena, and his two younger siblings. Elena, who completed high school but didn’t pursue further education, has always worked hard to provide for her children, often juggling multiple part-time jobs. Despite the challenges, she remains dedicated to her family, striving to offer them a better life than the one she had.

Elena’s marital status is divorced, a decision that came after a tumultuous marriage marked by her ex-partner’s struggle with alcohol and legal issues, including a recent arrest that further strained the family’s stability. This history of instability has left Elena wary of forming new romantic attachments, and she often feels isolated, lacking a strong support network to lean on during tough times.

Jordan is a boy who shows a lot of promise academically, especially for his age. However, Elena worries about the effects of their environment and her parenting style, particularly her reliance on strict discipline, which she fears might be too harsh at times. She is conscious of the cycle of harsh parenting she inherited from her own upbringing, which was marred by neglect and abuse.

Despite these challenges, Elena has never succumbed to alcohol or drug problems, a testament to her resilience and determination to provide a different life for Jordan and his siblings. She is, however, grappling with mental health issues, due to depression and anxiety stemming from years of stress and hardship. These issues have gone largely unaddressed, as Elena feels she lacks the resources and support to seek help.

The family’s dynamic is a complex one, with Elena doing her best to navigate the difficulties of raising three children on her own. The children’s ages span from Jordan’s 4 years to his youngest sibling, barely a year old, making the demands on Elena’s time and energy all the more challenging.

Elena’s greatest fear is that the legacy of hardship, coupled with her own struggles, might hinder her children’s chances for a happier, more stable life. She yearns for a supportive community and resources to help her address her mental health challenges and shift towards more positive parenting strategies.

Do NOT put in the values (below) for PRM first! Discuss in your group your predictions first.

Jordan: Values for PRM


Num children (chc_num): Elena has three children.
Child age (chdage): Jordan is 4 years old, falling into the 3-5 years category.
Child gender (chdgender): Jordan is male (M).
Marital status (PHH12a): Elena is divorced.
Education level (cgddegr): Elena completed high school (HS).
Has serious mental health or emotional problems (CRA17a): Yes, Elena struggles with mental health issues.
Any recent arrest/jail-time (CRA18a): No
Has active alcohol problems (CRA13a): No
Has active drug problems (CRA15a): No 
Has low social support (CRA45a): Yes, Elena feels isolated and lacks support.
Any history of abuse and neglect of the caregiver (CRA35a): Yes, Elena was subjected to neglect and abuse.
Harsh parenting (CTS-2) (PDS_DSPR): 70 because Elena uses strict discipline, indicating some level of harsh parenting.

ChatGPT powered conceptualization tool

Credits to Tan Yugin from the AI for Social Good lab

1. Link to tool

2. In case, the tool cannot work:

Case scenario

Presenting Issues at Start of Programme:

Mandated to counseling for theft.
Tense relationship with mother and current step-father.
Working part-time as event promoter and banquet server.
Returns home late, causing conflict.
Self-harm behaviors: slashing wrists, head-banging, punching walls/self.
Expressed suicidal ideation and poor self-esteem.

Relationship Dynamics:

Distant relationship with mother.
Heated arguments with step-father, affecting family environment.
Mother and step-father read C’s diary, revealing suicide thoughts.

Family Background:

C (18 years old, male, Chinese):

Lives in a rental flat with mother and step-father.
Met at-risk youths through classmates in Secondary 1
Lost interest in studies, dropped out of school in Secondary Two.
Ran away from home for a year after conflicts, stayed with friends or slept at void decks - Returned home recently
Started hanging out with colleagues after work, involved in underage drinking, engaged in theft to fund alcohol habit

Mother:

Currently in her second marriage.
Previous relationships included violent partners.
Accused of infidelity by previous husband, C developed mixed feelings towards mother as a result
Sided with step-father during quarrel about C’s behaviour.

Step-Father:

Previously incarcerated for assaults.
Abusive history.
Controlling, sets curfews, harsh on C.

Biological Father:

Alcohol addiction, financially unsupportive.
Violent towards family when C was in primary school.
Chased family out of home, family had no shelter for a time

Elder Sister:

Living independently with husband and children.

Output with “Risk” request

C (Client)

Physical Health: Engages in self-harm behaviors (slashing wrists, head-banging, punching walls/self), which indicates a risk to his physical well-being. Additionally, involvement in underage drinking poses further risks to his physical health.
Mental Health: Experiences poor self-esteem and has expressed suicidal ideation, indicating significant mental health challenges. The history of running away and engaging in theft also suggests underlying mental health issues.
Cognitive Ability: Dropping out of school in Secondary Two and losing interest in studies may reflect challenges in academic or cognitive engagement, possibly influenced by his social environment and mental health issues.
Motivation: His return home and engagement in part-time work could indicate some level of motivation towards stabilization, though his actions (theft, self-harm) suggest conflicting motivations or unmet needs.
Emotion: Exhibits signs of emotional distress, evidenced by self-harm and suicidal ideation. His tense relationship with family members also suggests emotional turmoil.
Behaviour: Engages in risky behaviors, including theft, underage drinking, and self-harm. His behavior has led to mandated counseling.
Reaction to Crisis: Running away from home and engaging in self-harm indicate maladaptive coping mechanisms in response to crisis situations.
Perception of Vulnerability: His actions and mental health issues suggest a heightened perception of vulnerability, though this may not be explicitly acknowledged.

Mother

Mental Health: Past relationships with violent partners and the accusation of infidelity by a previous husband may have impacted her mental health, though specific details are not provided.
Behaviour: Siding with the step-father during arguments and violating C’s privacy by reading his diary indicate problematic behaviors in handling family dynamics and C’s mental health issues.
Emotion: The dynamics described suggest potential emotional strain, though her emotional state is not explicitly detailed.

Step-Father

Physical Health: Previous incarceration for assaults indicates a history of physical aggression, though current physical health is not detailed.
Mental Health: A history of being abusive and controlling suggests potential underlying mental health or personality issues.
Behaviour: Exhibits controlling behavior and harsh treatment towards C, contributing to the tense family environment.

Biological Father

Physical Health: Alcohol addiction suggests negative impacts on his physical health.
Mental Health: The addiction and history of violence towards the family indicate significant mental health concerns.
Behaviour: Financially unsupportive and violent behavior has had a lasting impact on C and the family’s stability.

Elder Sister

The elder sister’s situation is mentioned but without specific challenges or risk factors related to C’s current situation, other than indicating a family member who has achieved a level of independence and stability.

Challenges and Risk Factors Relation to C

C’s behaviors and mental health issues are deeply intertwined with his family dynamics, including the history of violence, substance abuse, and emotional neglect or abuse from family members.
The step-father’s controlling behavior and history of violence contribute to the tense home environment, exacerbating C’s mental health challenges.
The mother’s past relationships and current siding with the step-father during conflicts may contribute to C’s feelings of isolation and mistrust within the family.
The biological father’s history of violence and alcohol addiction has likely had a long-term impact on C’s perception of family stability and safety, contributing to his current mental health and behavioral issues.
The elder sister’s independence might represent a contrast to C’s current situation, potentially offering a model for stability but also highlighting the dysfunction within his immediate family environment.

3. Polleverywhere for Values Choice Survey:

Twelve Responsible AI Values & definitions

Fairness: A fair NLP system treats all people equally. Developers of fair NLP systems ensure, as far as possible, that the system does not reinforce biases or stereotypes. A fair system works equally well for everyone independent of their race, gender, sexual orientation, and ability.
Privacy - An NLP system that respects people’s privacy implements strong privacy safeguards. Developers of privacy-preserving NLP systems minimize, as far as possible, the collection of sensitive data and ensure that the NLP system provides notice and asks for consent.
Sustainability: A sustainable NLP system preserves the environmental quality of current and future generations. Developers of sustainable NLP systems minimize, as far as possible, electricity use and reduce waste.
Inclusiveness: Inclusive NLP systems empower everyone and engage all people. Developers of inclusive AI systems consider, as far as possible, the needs of people who might otherwise be excluded or marginalized.
Safety: A safe NLP system performs reliably and safely. Developers of safe NLP systems implement strong safety measures. They anticipate and mitigate, as far as possible, physical, emotional, and psychological harms that the system might cause.
Social good: An NLP system that promotes social good supports, as far as possible, human wellbeing and flourishing, peace and happiness, and the creation of socio-economic opportunities.
Dignity: An NLP system that respects human dignity upholds the inherent worth of every individual. In addition to respective legislation, developers ensure that the system respects human rights and does not diminish human dignity.
Performance: A high-performing NLP system consistently produces good predictions, inferences or answers. Developers of high-performing NLP systems ensure, as far as possible, that the system’s results are useful, accurate and produced with minimal delay.
Accountability: An accountable NLP system has clear attributions of responsibilities and liability. Developers and operators of accountable AI systems are, as far as possible, held responsible for their impacts. An accountable system also implements mechanisms for appeal and recourse.
Transparency: A transparent NLP system produces decisions that people can understand. Developers of transparent AI systems ensure, as far as possible, that users can get insight into why and how a system made a decision or inference.
Human autonomy: An NLP system that respects people’s autonomy avoids reducing their agency. Developers of autonomy-preserving AI systems ensure, as far as possible, that the system provides choices to people and preserves or increases their control over their lives.
Solidarity: A solidary NLP system does not increase inequality and leaves no one behind. Developers of solidary AI systems ensure, as far as possible, that the prosperity as well as the burdens created by NLP are shared by all.