Info biomark

2,076 views

Published on

Information channels and biomarkers of disease

Published in: Technology, Health & Medicine
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
2,076
On SlideShare
0
From Embeds
0
Number of Embeds
1,533
Actions
Shares
0
Downloads
0
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide
  • Here recall: why SPSP / cits perspective is important. Not a posteriori reconstruction. Synchronic knolwledge building!
  • Floridi: constructionist, not representationalist. Knowledge is not a matter of hitting or finding, but of designing and constructing.reality is not a source but a resource for knowledge.
  • Info biomark

    1. 1. Information channels and biomarkers of disease Phyllis Illari and Federica Russo
    2. 2. Overview Why ‘exposomics’? A good case in the CitS / SPSP spirit In search of a concept of linking The failure of traditional production accounts Revisiting the causal literature in the light of new challenges Where to look next for a concept of linking The prospects of an informational approach 2
    3. 3. Why exposomics? 3
    4. 4. Environmental exposure and disease Traditional epidemiology Establish correlation between environmental factors and classes of disease Molecular epidemiology Measurement at molecular level Identify biomarkers of exposure, of early clinical changes, of disease 4
    5. 5. Finding the missing link Environmental factors ➯disease Chemicals in water / air ⤳ biomarkers of exposure ⤳ biomarkers of clinical changes ⤳ disease 5
    6. 6. From GWAS to EWAS The limits of genomics An expanded notion of exposure Internal and external exposure – the EXPOSOME Better understanding of disease mechanisms the MISSING LINK:how environmental exposure and disease are connected Go down to the molecular level Better prediction and health policy planning Earlier and more accurate prediction Identification of areas for public health intervention 6
    7. 7. Goals Looking for biomarkers of the process of evolution of the disease The meeting-in-the-middle methodology: matching key evolution stages 7 Come to understand how environmental exposure is linked to the disease We know a lot about the system, but a lot is still unknown We seek many small causes, with small effects, and large interaction effects We expect widely different factors to be causes, including e.g. social and chemical factors Use of new omics technologies allows a comprehensive approach, but generates vast amounts of data Challenges Upshot Exposomicslooks for tiny, difficult-to-find causal links in the middle of a mess
    8. 8. What is this linking?
    9. 9. What we want Nail down what linkingisand how we reason about it Tension: general but not vacuous A thin and versatile metaphysics – and broadly realist A widely applicable, general metaphysics – and informative 9
    10. 10. Where to find linking? Accounts of difference-making Probabilistic, counterfactual, manipulationist, … Establish thatC causes E Accounts of production Processes, mechanisms, CPD, … Establish howC causes E 10
    11. 11. TRADITIONAL ACCOUNTS OF PRODUCTION FAIL 11
    12. 12. Physical processes Tied to description of processes in physics, only one type of linking Mechanisms Give organization of parts and their interactions, course-grained linking Capacities, Powers, Dispositions Give modal properties of some parts of mechanisms, not linking 12
    13. 13. Purely physical quantity Physical processes Tied to description of processes in physics, only one type of linking Mechanisms Give organization of parts and their interactions, course-grained linking Capacities, Powers, Dispositions Give modal properties of some parts of mechanisms, not linking
    14. 14. Processes Discriminate between causal and non-causal physical processes Salmon-Dowe Exchange of conserved quantities Billiard balls colliding Boniolo et al Exchange between extensive quantities Electric heating making environment hotter Tied to description of processes in physics, only one type of linking 15
    15. 15. Physical processes Tied to description of processes in physics, only one type of linking Mechanisms Give organization of parts and their interactions, course- grained linking Capacities, Powers, Dispositions Give modal properties of some parts of mechanisms, not linking
    16. 16. Mechanisms Illari& Williamson proposed consensus: “Entities and activities organised in such a way of being responsible for a phenomenon” Key elements of mechanistic explanation: Identification of the phenomenon Identification of entities and activities involved Identification of the organisation Give organization of parts and their interactions, course-grained linking 17
    17. 17. Physical processes Tied to description of processes in physics, only one type of linking Mechanisms Give organization of parts and their interactions, course-grained linking Capacities, Powers, Dispositio ns Give modal properties of parts of mechanisms, not linking
    18. 18. Capacities, Powers, Dispositions Cartwright & Pemberton, Mumford &Anjum, Bird Cope with (scientific) situations lacking universal, exceptionless laws Hang on to realist intuitions CPDs are properties of a particular thing or system Things-with-capacities licence specific capacity claim Modal character of capacities Give properties of some parts of mechanisms, not linking 19
    19. 19. Recap Traditional accounts of production seek: Realism: what is causality? Science-first: what does science say causality is? Exposomics requires something: Fine-grained Very general to apply in multiple cases and to link profoundly inhomogeneous causal factors Something that can exist and potentially be found in such a mess of interaction 20
    20. 20. How to trace such linking?
    21. 21. THE PROSPECTS OF AN INFORMATIONAL APPROACH 22
    22. 22. Picking up signal While classical statistical models to analyzing -omics data serve the purpose of identifying signals and separating them from noise, little has been done in chronic diseases to model time into the exposure-biomarker-disease continuum. Recently, SandroGalea has proposed to generalise mathematical approaches initially designed to study infectious disease to chronic diseases epidemiology. In that setting, biomarkers should not be assessed synchronically, as usually done, but diachronically and even ideally their full evolution along time should be considered. [Vineis and Chadeau-Hyam 2011] From these two parallel analyses [statistical analyses], we obtained lists of putative markers of (i) the disease outcome, and (ii) exposure. These were compared in a second step in order to identify possible intersecting signals, therefore defining potential intermediate biomarkers. [Chadeau-Hyam et al 201] 23
    23. 23. Why information Gives a way of describing reality, of any kind. Information transmission can be tracked, measured, described precisely. See e.g.: biological information A thin secret connexion In line with Anscombian pluralism Guides interpretation of multiple data sources Helps with knowledge construction, conceptualisation of link, of picking up signal 24
    24. 24. Informational thinking helps with knowledge construction Information is like a branch floating down a river Scientific study of processes, mechanisms, CPDs Define the banks, find the currents, intercept position of branch, … The river is there, but we shape the banks We try to track the tree down to the wood factory, where it can be processed 25
    25. 25. Mechanisms as information channels Reconcile informational approach to successful existing work on mechanisms Mechanisms give course-grained linking They help track linking 26
    26. 26. Why informational thinking helps: the inhomogeneity argument ⁇Plausible links Social ⤳ Biological Big ⤳ Small (and viceversa) ☝Exposomics Environmental exposure (B) ⤳ molecular changes (s) ‼ Restore homogeneity: biological measurements ⁇Problem isn’t solved Choose exact level, links hard to interpret Reinterpret micro links at macro level ☝Information flowing via biomarkers: Interpretation of many pieces of puzzle Reconstruction of information flow Biomarkers analyses + statistics + biological theory + … ⁇Plausible links Not via homogeneity But via reconstruction of continuous linking 27
    27. 27. WHAT LINKING IS NOT 28
    28. 28. Avoid category mistakes Why not counterfactuals? Isn’t causation counterfactual dependence? Counterfactuals help with reasoning Counterfactual dependence helps establish difference-making, not production 29
    29. 29. SUM UP AND CONCLUDE 30
    30. 30. Why exposomics? An emerging, interesting, challenging area of research driving change Modelling disease evolution Conceptualisation of exposure A field to test the possibility of philosophy to help with conceptual engineering Do traditional accounts help with new challenges? 31
    31. 31. Why information Referenced in scientific practice Picking up signal More than an entity in our ontology A thin secret connexion A way of thinking Knowledge construction and data interpretation 32
    32. 32. Why information for exposomics? Exposomics forces us to reconsider existing approaches to causality: shows the importance of understanding causal linking… …while showing that existing accounts of productionhelp track linking Information as an answer is Real and referenced in scientific practice, General and fine-grained enough, Helps with tracking linking, by reconstructing the flow of signaling 33

    ×