- With Ellie Dillon (Aalto-yliopisto), Eeva Vilkkumaa (Aalto-yliopisto) & Finnish Cancer Registry
- Ongoing academic work – results will be published later
DESCRIPTION: We develop an optimization algorithm to optimize cancer prevalence reduction per required colonoscopies. The strategies consist of selecting invited segments and optimal FIT test thresholds for them.
The algorithm accounts for:
- Multiple objectives (prevalence, number of colonoscopies, direct costs)
- Resource constraints
- Differences between population segments (age, sex)
- Imperfect information and event probabilities for e.g. false test results, adverse events, polypectomies etc.
APPLICABILITY: The resulting algorithm and framework can be used to optimize different types of screening strategies in health care, maintenance etc or potentially any periodic decision making problem where results from last period affect best strategies in the next one.
Due to complexity of the problem, the algorithm is computationally heavy. However, this reflects mostly in time required to compute the answers, not so much in memory requirements.
TOOLS: