Solutions and Services
Applying well-established AI techniques
to complex problems.
Solutions
JOB AIDS
We create job aids for researchers, frontline healthcare workers and programme designers. These digital & paper-based tools are backed by machine learning and weighted scorecards to assess risk and triage patients in real-time.
PATIENT ARCHETYPES
We develop archetypes or personas that can be used to pair patients with the most suitable program interventions. By going beyond key populations and segmenting more granularly (e.g. “Disillusioned Discloser”, “Frequently Late”, “Facility Shopper”, “Lone Ranger”, “Working Youth”), each with distinct risk factors, the archetypes enable targeted, tailored interventions and other forms of differentiated care.
RISK SCORING
We use predictive analytics and alternative data to model the behaviour and outcomes of those who find themselves without access to traditional care and services. Understanding how risk varies across a population allow for more precise interventions and for providers to target individuals most in need.
Whether building credit scores for “unscorables” or determining the likelihood a patient returns to care in a community clinic, many private sector techniques can amplify development programs by delivering key insights and better decisioning.
COHORT SEGMENTATION
Using unsupervised learning and advanced clustering we can find natural groups and connected traits in sub-populations. We use these to see how the characteristics correlate and influence outcomes our partners aim to impact.
We can then profile the risk of individuals so that expert users can refactor programmes or point of care decision support. Another use case would be to identify locations to
Services
Palindrome can work with you to absorb the value of these solutions in your organisation. Using our experience in project support and management consulting,
we can help in 6 ways:
.
Feasibility Study - Expected benefits of machine learning techniques
Many organisations are uncertain whether they “are ready” for Data Science techniques like AI and ML, and whether the hype will lead to tangible benefits in their organisation. Palindrome will work with partners to assess their readiness and the projected quantifiable benefits.
Data assessment and preparation
Curating, cleaning and encoding in-house data is often the most involved activity for many data analytics projects – and something partners can struggle to find time for in-house. This activity also involves augmenting in-house data with external sources and additional features that add new signals for the models to leverage
Customised Model Building
Palindrome will build custom predictive models around key outcomes and business metrics. With an array of advanced modelling techniques, Palindrome will find the optimal scientific approach to fit the partners business goals and operating requirements.
Field Testing
A key step for any model is to have its outputs tested in the operating environment. Not only can we test the model’s efficacy against control interventions, but we’ll gain insights from extensive user-testing that help the design of the intervention and model.We then review what the organisational gaps and barriers are to operationalizing the work.
Risk Management and Regulatory Compliance
Palindrome will begin by understanding and measuring the risks and exposure. As regulations are constantly shifting and not broadly understood, Palindrome will work to understand the specific needs for model governance, and consumer consent & privacy documentation.
Operationalization
Once the field testing is satisfactory, and the business case for operationalizing the model has been approved, Palindrome will work with our partners to establish process and change management procedures for the adoption of the model into the daily business operations.
