The document describes a model developed using multi-level Petri nets to simulate the spread of antibiotic resistance in microbiota and human populations. The model represents microbiota as a middle level, with bacterial cells that can be resistant or non-resistant, and human hosts as a top level that can experience mild or severe resistance based on the microbiota resistance levels. The model is used to experiment with how different antibiotic administration protocols affect microbiota and population resistance levels over time. The authors conclude the model provides flexibility to explore computational roles in addressing antibiotic resistance and plan to increase complexity and incorporate real data in future work.

1. Using Multi-level
Petri Nets Models to
Simulate Microbiota
Resistance to
Antibiotics
R. Bardini, G. Politano, A. Benso and S. Di Carlo
2017 IEEE International Conference on Bioinformatics an Biomedicine.The
Westin at Crown Center Hotel, Kansas City, MO, USA, Nov. 13-16, 2017

2. MOTIVATIONS
The spread of antibiotic resistance: an imminent threat
[CDC, 2017]

3. MOTIVATIONS
The spread of antibiotic resistance: an imminent threat
[CDC, 2017]

4. Approaching the problem
The nets-within-nets formalism
The model
Experimental designs
Conclusions and future work
OUTLINE

5. FDF
SYSTEMIC INTERVENTIONS
APPROACHINGTHE PROBLEM

6. FDF
SYSTEMIC INTERVENTIONS
APPROACHINGTHE PROBLEM
POLICY
MAKING

7. FDF
SYSTEMIC INTERVENTIONS
APPROACHINGTHE PROBLEM
POLICY
MAKING
Clinical Innovations

8. FDF
SYSTEMIC INTERVENTIONS
APPROACHINGTHE PROBLEM
POLICY
MAKING
Clinical Innovations
TREATMENT
PROTOCOLS

9. FDF
SYSTEMIC INTERVENTIONS
APPROACHINGTHE PROBLEM
POLICY
MAKING
Clinical Innovations
TREATMENT
PROTOCOLS
BACTERIAL
REINTEGRATION

10. FDF
SYSTEMIC INTERVENTIONS
APPROACHINGTHE PROBLEM
POLICY
MAKING
Clinical Innovations
TREATMENT
PROTOCOLS
BACTERIAL
REINTEGRATION
ALTERNATIVE
TREATMENTS

11. FDF
SYSTEMIC INTERVENTIONS
APPROACHINGTHE PROBLEM
POLICY
MAKING
Clinical Innovations
TREATMENT
PROTOCOLS
BACTERIAL
REINTEGRATION
ALTERNATIVE
TREATMENTS
Analysis

12. FDF
SYSTEMIC INTERVENTIONS
APPROACHINGTHE PROBLEM
POLICY
MAKING
Clinical Innovations
TREATMENT
PROTOCOLS
BACTERIAL
REINTEGRATION
ALTERNATIVE
TREATMENTS
PredictionAnalysis

13. FDF
SYSTEMIC INTERVENTIONS
APPROACHINGTHE PROBLEM
POLICY
MAKING
Clinical Innovations
TREATMENT
PROTOCOLS
BACTERIAL
REINTEGRATION
ALTERNATIVE
TREATMENTS
Prediction DesignAnalysis

14. APPROACHINGTHE PROBLEM
The roles of computational models
Prediction DesignAnalysis
Understand the
complexity
Develop strategies
Implement
good practices

15. NETS-WITHIN-NETS
An high-level, object-oriented Petri Nets formalism

16. NETS-WITHIN-NETS
An high-level, object-oriented Petri Nets formalism

17. NETS-WITHIN-NETS
An high-level, object-oriented Petri Nets formalism

18. NETS-WITHIN-NETS
An high-level, object-oriented Petri Nets formalism

19. NETS-WITHIN-NETS
An high-level, object-oriented Petri Nets formalism

20. NETS-WITHIN-NETS
An high-level, object-oriented Petri Nets formalism

21. THE MODEL
An overview of the general architecture

22. THE MODEL
Middle level: the microbiota
ddddas
BACTERIAL CELLS CAN BE IN TWO
STATES: RESISTANCE AND NON
RESISTANCE

23. THE MODEL
Middle level: the microbiota
POPULATION DYNAMICS EMERGE FROM
GROWTH RATES AND RESOURCE
AVAILABILITY

24. THE MODEL
Middle level: the microbiota
NON-RESISTANT CELLS CAN PASS TO
THE RESISTANT STATE THROUGH
HORIZONTAL GENETIC TRANSFER
MECHANISMS

25. THE MODEL
Middle level: the microbiota
ANTIBIOTIC ADMINISTRATION EVENTS
CAN BE TRIGGERED

26. THE MODEL
Middle level: the microbiota
BACTERIAL REINTEGRATION ACTIONS
CAN TAKE PLACE DURING ANTIBIOTIC
TREATMENT

27. THE MODEL
Middle level: the microbiota
A POINT PREVALENCE SCORE FOR
RESISTANCE IS DYNAMICALLY
COMPUTED
# resistant bacterial cell / # total bacterial
Point Prevalence Score (PPS)

28. THE MODEL
Top level: the human hosts
EACH HOST HAS A MICROBIOTA WHICH
CAN LIVE IN THREE STATES: HEALTH,
MILD RESISTANCE AND SEVERE
RESISTANCE

29. THE MODEL
Top level: the human hosts
ONE OR MORE MICROBIOTA CAN LIVE
INTO THE TOP LEVEL NET

30. THE MODEL
Top level: the human hosts
THE RESISTANCE STATE VARIES ACCORDING
TO THE RESISTANCE POINT PREVALENCE
SCORE FOR THE MICROBIOTA

31. THE MODEL
Top level: the human hosts
THE TOP LEVEL CAN OPERATE
DIFFERENT PROTOCOLS OVER THE
MICROBIOTA

32. THE MODEL
Top level: the human hosts
IT IS POSSIBLE TO TRACK
VARIABLES OF INTEREST ALONG THE
SIMULATION TIME

33. EXPERIMENTAL DESIGN I
How do different antibiotic administration protocols affect the resistance state of a microbiota?
# resistant bacterial cell / # total bacterial
Point Prevalence Score (PPS)

34. EXPERIMENTAL DESIGN I
How do different antibiotic administration protocols affect the resistance state of a microbiota?
# resistant bacterial cell / # total bacterial
Point Prevalence Score (PPS)
0
20
40
60
80
100
1
101
201
301
401
501
601
701
801
901
1001
1101
1201
1301
1401
PPS
Simulation time
0
20
40
60
80
100
1
101
201
301
401
501
601
701
801
901
1001
1101
1201
1301
1401
PPS
Simulation time
0
20
40
60
80
100
1
106
211
316
421
526
631
736
841
946
1051
1156
1261
1366
1471
PPS
Simulation time
No treatment No treatment
0
20
40
60
80
100
APPS
NO TREATMENT TRADITIONAL TREATMENT INNOVATIVE PROTOCOL
a b c
d
i
e
hg
f pre-treat dose 1 + prev dose 2
0
20
40
60
80
100
APPS
pre-treat. dose 1 dose 2
0
20
40
60
80
100
APPS

35. EXPERIMENTAL DESIGN II
How do different antibiotic administration protocols affect the resistance state of a human population?
Simulation time for severe resistance onset
Resistance Onset Time

36. EXPERIMENTAL DESIGN II
How do different antibiotic administration protocols affect the resistance state of a human population?
Simulation time for severe resistance onset
Resistance Onset Time
SINGLE ANTIBIOTIC DOSE
a
SINGLE ANTIBIOTIC DOSE + MICROB. INTEGR.
b
c d
0
10
20
30
40
50
1
15
29
43
57
71
85
99
113
127
141
155
169
183
197
211
225
239
Populationnumerosity
Simulation Time
0
10
20
30
40
50
1
21
41
61
81
101
121
141
161
181
201
221
241
261
281
301
321
341
Populationnumerosity
Simulation Time

37. • Exploration of possible roles for computational models in
solving a systemic problem such as the spread of
antibiotic resistance
CONCLUSIONSAND FUTURE
WORK

38. • Exploration of possible roles for computational models in
solving a systemic problem such as the spread of antibiotic
resistance
• High level of abstraction, high flexibility
CONCLUSIONSAND FUTURE
WORK

39. • Exploration of possible roles for computational models in
solving a systemic problem such as the spread of antibiotic
resistance
• High level of abstraction, high flexibility
• More complexity to be included at the microbiota level
(diversity, interactions between species, plasticity,…) and at
the human population level (possible interactions between
hosts, …)
CONCLUSIONSAND FUTURE
WORK

40. • Exploration of possible roles for computational models in
solving a systemic problem such as the spread of antibiotic
resistance
• High level of abstraction, high flexibility
• More complexity to be included at the microbiota level
(diversity, interactions between species, plasticity,…) and at the
human population level (possible interactions between hosts,
…)
• Working with first-hand and quantitative data (combining
hypothesis-driven and data-driven approaches)
CONCLUSIONSAND FUTURE
WORK