Cambridge Humanae Reliability via Human Understanding
Cambridge Humanae Nicholas Kirk Andrea Stefani Alex Kölpin Stefan Zeidler Masoud Kamali Psychology Investor CEO Former scientist at Fraunhofer ESK Professional training in Psychometrics (University of Cambridge) M.Sc. major Artificial Intelligence (Technical University of Munich) Business Investor Rockstart AI Investor
30% more stress in the last 30 years
Effect on sectors Banking Venture capital Consulting Safety Corporate HR Aviation
Use Cases Improve selection Improve operations Reduce training dropouts Reduce training length
Clear problem Image damage Stress causes sick leave Organization disruption Financial costs: 543MM loss/year
Known from [1] Nicholas Hubert Kirk and Karmela Arbanasic. Reducing inoperative days of soldiers under stress: A novel approach. In Proceedings of the 59th Annual Conference of the International Military Testing Association, Bern, Switzerland, 2017. [2] Karmela Arbanasic and Nicholas Hubert Kirk. Understanding resilience by relating stimuli optimality to schema constructs. In Proceedings of the 59th Annual Conference of the International Military Testing Association, Bern, Switzerland, 2017.
IAMPS 2018 PERSONA: a Psychometric Monitoring Instrument for High-stress Individuals pgyh5 https://psyarxiv.com/
Problem Individuals in high-risk professions are: Overly stressed on duty and non treated timely Non sufficiently able to cope with stress both inside and outside their workplace Need for a time-conscious assessment tool, to be applied during selection and monitoring of personnel in sensitive roles, which blends multiple clinical-level and personality traits, for behaviour prediction
https://psyarxiv.com/pgyh5
Theoretical leverage Need for stimuli Pearson s categorization of stimuli type and preferences (Pearson, 1970) Cognitive schemas Fowler s model of self-schema and world-schema (Fowler et al., 2006) Coping mechanisms Hierarchical model of coping mechanisms (Bond & Vaillant, 1986) (DSM-III)
Our model Stress as over- & under- stimulation, (i.e. deviation from optimal stimulation level (OSL) (Hebb, 1955)) Cognitive schemas as autobiographical memory recall, mapped on positive-negative and important-unimportant scales Coping mechanisms as reality distortion, based on the Hierarchical model of coping mechanisms (Bond & Vaillant, 1986) (DSM-III)
Test internals 80 initial items PHASE I: 8 items PHASE II: 32 items PHASE III: 40 items (saturated more on the more psychotic dimensions)
Item construction examples PI.EC.P. I feel the need to talk to curious people PII.SP.IS.P. I feel negative when I think of people who showed me empathy PII.SI.IC.O. I rarely think of some particular events which brought me new unexpected thoughts PII.OP.IS.P. I feel positive when I think about people who pay attention to the emotions of others PIII.IB.2. I appreciate the bad events that happened to me, because they made me happier than I would ever be without them PIII.IM.3. When somebody seems as honest person to me, s/he usually later disappoints me with his/her behavior PIII.DA.1. If somebody bothers me, they deserve a strong reaction
Evaluation test set 67 individuals (general population, English Test, Europe) 16 individuals (clinical population, Italian Test, S. Chiara Hospital, Pisa) Certified diagnoses included major depressive disorder, panic disorder and borderline personality disorder. We consider the triggered clinical state as an extreme condition potentially brought by stressful conditions
Hierarchy of Stress Defense Mech: DSM-III Mature Humor, suppression, sublimation Neurotic Repression, displacement, rationalization Immature Passive aggression, acting out, dissociation Psychotic Denial, distortion
Reliability Adequate levels of internal consistency both at a test-level and at a scale-level (Cronbach Alpha-based assessment) PHASE I, good consistency (0.74) PHASE II, very good consistency (0.88) PHASE III, slightly insufficient (0.68) (too much heterogeneity) Factor structure did not fully confirm the three-phase architecture, but it sufficiently divided PHASE I from PHASE III, therefore justifying the separation rationale
Purification via Machine Learning: PERSONA-16 Predictive validity towards generic psychopathological state classification >97%
Conclusions Predictive validity > 97% Decent factor structure Decent internal consistency (c-alpha) Separation between PHASE I and PHASE III The test has passed a preliminary validation. We would like to expand to multiple domains and multiple uses, given the wide theoretical framework.
Call for action The paper in the link contains the purified PERSONA-16. The validated architecture can be a structure for multiple implementations. Feel free to validate further & use the instrument (reference us!) Engage with us for collaborations (email us!) https://psyarxiv.com/pgyh5 kirk@humanae.xyz
Backup
Possible outcomes of the test, given estimates of test constructs Cognitive schemas Production function Coping mechanisms Level of psychopathology associated with given coping mechanism Positive Destructive Mature Low level of psychopathology Negative Constructive Mature Low level of psychopathology Negative Destructive Immature High level of psychopathology Positive Destructive Immature High level of psychopathology Negative Constructive Immature High level of psychopathology
Psychopathological implications, given the estimates of cognitive schemas and coping mechanisms Self-schema World-schema Coping mechanism Psychological manifestation Positive Neutral Information bias Extreme optimism, possible lack of critical thinking, possible narcissism Neutral Neutral Information reorganization Extreme rationalization, emotional isolation Incoherent Incoherent Information modification Incoherent behavior Neutral or negative Negative Indirect action Anxiety, depressive or stress-related disorders, suicidal thoughts Negative Negative Direct action Depressive or stress-related disorders, violent behavior, suicidal thoughts Extremely negative Extremely negative Reality distortion Personality disorders, psychotic disorders, sociopathic tendencies