martha flora- d-ai-versifying health care
TRANSCRIPT
D-AI-versifying health care
CASE STUDY
MARTHA FLORA – SEEDLINK March 21st 2017
INEKE PET – PIET HEIN VAN DAM HR Tech - London
www.seedlinktech.com
Confidential for Shanghai United Family Hospital (not for distribution)
Challenge: Predicting Resident Service Behavior
How do we know if our staff gives our
residents the right service experience?
Vision
Martha Flora’s vision is that care is based on the
relationship between people.
The quality of care is defined by the capability of the
care giver to build a relationship with the patient and
his family.
The medical aspect is arranged perfectly, but the focus is
based on the welfare of the residents and their family.
Halo Effect Similarity Bias
Solution: Make data-driven decisions with artificial intelligence
Unlike assessment centers, personality tests and human interviewers
Seedlink uses Natural Language AI on behavioral traits to make data-
driven recommendations that get better over time
Seedlink methodology is build on 20+ years
research Language correlates with human behavior1
Multi-Modal predictive technology2, 3, 4
1. Pennebaker, J. & King, L. (1999). Linguistic styles : Language use as an individual difference.
2. Pennebaker JW, Mehl MR, Niederhoffer KG. Psychological aspects of natural language use: Our words, our selves
3. Verhoeven, B. & Daelemans, W. (2014). CLiPS Stylometry Investigation (CSI) corpus: A Dutch corpus for the detection of age,
gender, personality, sentiment and deception in text.
4. Gao, R., Hao, B., Bai, S., Li, L., Li, A. & Zhu, T. (2013). Improving user profile with personality traits predicted from social media
content. Conference on Recommender Systems.
1
• Collect
language data
from employees
and understand
their behaviors
2
• Artificial
intelligence
software builds
your predictive
company
cultural
benchmark
3
• Get insights in the
behaviors of
success
• Discover skills gaps
for development
• Hire better by
predicting who fits
your culture best
Easy 3-step approach to build predictive model
Model takes long term performance data as start point
Algorithms study language of high performers as unique identifier
Methodology:
1400 candidates’ from various backgrounds are rated on
self-monitoring and ability to adapt for Alzheimer’s care
facility.
3 years collaboration
Results:
- Over 70% predictive value on high performing hires
- Over 90% match for rejects
- Recommendations had over 5-fold greater chance of
success
How do our algorithms compare to human
selected traits?
Seedlink PeopleInsights
Seedlink test outperforms all methods on reliability & accuracyPeopleInsights Enterprise model shows 0.82 predictive reliability to job performance1
Personality tests are easily manipulated and results change based on context 2,3
1. Based on initial client case studies for experienced hiring Seedlink
2. Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology:
practical and theoretical implications of 85 years of research findings.
3. Rynes, S. L., Colbert, A. E., & Brown, K. G. (2002). Hr professionals' beliefs about effective human resource
practices: correspondence between research and practice
0.82
Thank you for your attention!
INEKE PET [email protected]
PIET HEIN VAN DAM [email protected]
@pietheinvandam
www.seedlinktech.com
Confidential for Shanghai United Family Hospital (not for distribution)