scholarly journals Incorporating Behavioral Constraints in Online AI Systems

2019 ◽  
Vol 33 ◽  
pp. 3-11 ◽  
Author(s):  
Avinash Balakrishnan ◽  
Djallel Bouneffouf ◽  
Nicholas Mattei ◽  
Francesca Rossi
Keyword(s):  
Real World ◽  
Upper Bound ◽  
Daily Life ◽  
Exploration And Exploitation ◽  
Real World Data ◽  
Thompson Sampling ◽  
Behavioral Constraints ◽  
Online Setting ◽  
Additional Constraints

AI systems that learn through reward feedback about the actions they take are increasingly deployed in domains that have significant impact on our daily life. However, in many cases the online rewards should not be the only guiding criteria, as there are additional constraints and/or priorities imposed by regulations, values, preferences, or ethical principles. We detail a novel online agent that learns a set of behavioral constraints by observation and uses these learned constraints as a guide when making decisions in an online setting while still being reactive to reward feedback. To define this agent, we propose to adopt a novel extension to the classical contextual multi-armed bandit setting and we provide a new algorithm called Behavior Constrained Thompson Sampling (BCTS) that allows for online learning while obeying exogenous constraints. Our agent learns a constrained policy that implements the observed behavioral constraints demonstrated by a teacher agent, and then uses this constrained policy to guide the reward-based online exploration and exploitation. We characterize the upper bound on the expected regret of the contextual bandit algorithm that underlies our agent and provide a case study with real world data in two application domains. Our experiments show that the designed agent is able to act within the set of behavior constraints without significantly degrading its overall reward performance.

scholarly journals Using Contextual Bandits with Behavioral Constraints for Constrained Online Movie Recommendation

2018 ◽  
Author(s):  
Avinash Balakrishnan ◽  
Djallel Bouneffouf ◽  
Nicholas Mattei ◽  
Francesca Rossi
Keyword(s):  
Daily Life ◽  
Ethical Principles ◽  
Online System ◽  
Interactive Interface ◽  
Behavioral Constraints ◽  
Online Setting ◽  
Movie Recommendation ◽  
Overall Performance ◽  
Additional Constraints

AI systems that learn through reward feedback about the actions they take are increasingly deployed in domains that have significant impact on our daily life. In many cases the rewards should not be the only guiding criteria, as there are additional constraints and/or priorities imposed by regulations, values, preferences, or ethical principles. We detail a novel online system, based on an extension of the contextual bandits framework, that learns a set of behavioral constraints by observation and uses these constraints as a guide when making decisions in an online setting while still being reactive to reward feedback. In addition, our system can highlight features of the context which are more predicted to be more rewarding and/or are in line with the behavioral constraints.  We demonstrate the system by building an interactive interface for an online movie recommendation agent and show that our system is able to act within a set of behavior constraints without significantly degrading overall performance.

scholarly journals Development of an Innovative Real-World Evidence Registry for the Herpes Simplex Virus: Case Study

10.2196/16933 ◽  
2020 ◽  
Vol 3 (1) ◽  
pp. e16933 ◽  
Author(s):  
Michelle Helena van Velthoven ◽  
Ching Lam ◽  
Caroline de Cock ◽  
Terese Stenfors ◽  
Hassan Chaudhury ◽  
...  
Keyword(s):  
Herpes Simplex Virus ◽  
Herpes Simplex ◽  
Real World ◽  
Physical Symptoms ◽  
Use Cases ◽  
Real World Data ◽  
World Data ◽  
Real World Evidence ◽  
Simplex Virus

Background Infection with the herpes simplex virus (HSV) is common but not well understood. Furthermore, there remains a social stigma surrounding HSV that can have psychosocial implications for those infected. Despite many patients infected with HSV experiencing mild-to-severe physical symptoms, only one subeffective treatment is available. A registry collecting real-world data reported by individuals potentially infected with HSV could help patients to better understand and manage their condition. Objective This study aimed to report on the development of a registry to collect real-world data reported by people who might be infected with HSV. Methods A case study design was selected as it provides a systematic and in-depth approach to investigating the planning phase of the registry. The case study followed seven stages: plan, design, prepare, collect, analyze, create, and share. We carried out semistructured interviews with experts, which were thematically analyzed and used to build use cases for the proposed registry. These use cases will be used to generate detailed models of how a real-world evidence registry might be perceived and used by different users. Results The following key themes were identified in the interviews: (1) stigma and anonymity, (2) selection bias, (3) understanding treatment and outcome gaps, (4) lifestyle factors, (5) individualized versus population-level data, and (6) severe complications of HSV. We developed use cases for different types of users of the registry, including individuals with HSV, members of the public, researchers, and clinicians. Conclusions This case study revealed key considerations and insights for the development of an appropriate registry to collect real-world data reported by people who might be infected with HSV. Further development and testing of the registry with different users is required. The registry must also be evaluated for the feasibility and effectiveness of collecting data to support symptom management. This registry has the potential to contribute to the development of vaccines and treatments and provide insights into the impact of HSV on other conditions.

Security of Railway Infrastructures

2012 ◽  
pp. 355-379
Author(s):  
A. Di Febbraro ◽  
F. Papa ◽  
N. Sacco
Keyword(s):  
Risk Analysis ◽  
Real World ◽  
Analysis Tool ◽  
Security Risk ◽  
Input Output ◽  
Real World Data ◽  
Security Problem ◽  
Output Characteristics ◽  
General Architecture

The chapter is organized as follows: In section 1, the basic definitions of the security risk analysis and the characteristics of the railway security problem are introduced, and a bibliography review is reported. Then, in section 2, the general architecture for designing a security risk analysis tool is presented, focusing on the relevant specifications, and on the input/output characteristics. Therefore, in section 3, with the aim of pointing out the characteristics of the presented architecture, an explicative case study is defined based on real world data coming from Italian railways. Finally, some conclusions and remarks are discussed in chapter 4.

scholarly journals Developing Markov Models From Real-World Data: A Case Study of Heart Failure Modeling Using Administrative Data

2020 ◽  
Vol 23 (6) ◽  
pp. 743-750
Author(s):  
Praveen Thokala ◽  
Peter Dodd ◽  
Hassan Baalbaki ◽  
Alan Brennan ◽  
Simon Dixon ◽  
...  
Keyword(s):  
Heart Failure ◽  
Administrative Data ◽  
Real World ◽  
Markov Models ◽  
Real World Data ◽  
World Data ◽  
Failure Modeling

scholarly journals Proportional ranking in primary elections: A case study

2022 ◽  
pp. 135406882110667
Author(s):  
Ariel Rosenfeld ◽  
Ehud Shapiro ◽  
Nimrod Talmon
Keyword(s):  
Cluster Analysis ◽  
Political Parties ◽  
Real World ◽  
Primary Elections ◽  
Real World Data ◽  
World Data ◽  
Primary Election ◽  
Aggregation Methods ◽  
Voter Groups

Many democratic political parties hold primary elections, which nicely reflects their democratic nature and promote, among other things, the democratic value of inclusiveness. However, the methods currently used for holding such primary elections may not be the most suitable, especially if some form of proportional ranking is desired. In this paper, we compare different algorithmic methods for holding primaries (i.e., different aggregation methods for voters’ ballots) by evaluating the degree of proportional ranking that is achieved by each of them using real-world data. In particular, we compare six different algorithms by analyzing real-world data from a recent primary election conducted by the Israeli Democratit party. Technically, we analyze unique voter data and evaluate the proportionality achieved by means of cluster analysis, aiming at pinpointing the representation that is granted to different voter groups under each of the algorithmic methods considered. Our finding suggest that, contrary to the most-prominent primaries algorithm used (i.e., Approval), other methods such as Sequential Proportional Approval or Phragmen can bring about better proportional ranking and thus may be better suited for primary elections in practice.

VP49 Real-world Evidence For Economic Evaluation Of Medical Devices

2019 ◽  
Vol 35 (S1) ◽  
pp. 87-88
Author(s):  
Benedetta Pongiglione ◽  
Aleksandra Torbica
Keyword(s):  
Medical Devices ◽  
Real World ◽  
Epidemiological Data ◽  
Evidence Base ◽  
National Registry ◽  
Economic Research ◽  
Data Mapping ◽  
Real World Data ◽  
World Data

IntroductionRandomized controlled trials (RCTs) are considered the gold standard in the hierarchy of research designs for evaluating the efficacy and safety of a treatment intervention. The low external validity of RCTs and the general shortage of clinical evidence available to support the use of many medical devices have emphasized the necessity for exploring the use of real-world data (RWD) as a complementary source to RCTs data for establishing a more robust evidence base on the effectiveness of medical devices. The aim of the present project is to assess in a comprehensive way the existing sources of real world data on medical devices in Europe.MethodsThe guidelines to the mapping exercise have been outlined in a research protocol. First, all national relevant sources (e.g. website of Ministry of Health, national institutions, research bodies) are screened, both in local language and English. Second, we perform a systematic search on PubMed using a set of key words for each case study, adapted to each country setting. Finally, we seek advice from key actors in the field of the device and clinical conditions, such as manufacturers or clinicians.ResultsInformation on existing sources of RWD for each case studies are provided in a template including details on the key features of the source (e.g. data producer, data collection period, sample size, study design, geographical coverage) and the main content of the dataset, distinguishing socio-demographic information, clinical and epidemiological data, data on resource use and health outcomes. The data mapping includes all countries of the project participants, i.e. Italy, UK, Netherlands, Switzerland, Germany, Hungary, and we enlarge the scope of our mapping including other countries: Spain, France, Denmark, Finland, Sweden, Poland and Hungary as well as international databases at pan-EU level. The number of available sources of RWD and their quality vary depending on case study and across countries. For example, in the case of orthopaedics, many countries have a national registry and administrative data, such as hospital discharge, contain useful information, although not as detailed. When a registry is not available, it is often the case that more observational studies are available; this occurs for example in France.ConclusionsIn this work we shows the importance of RWE and map in an accurate and comprehensive way which source of RWD are currently available and to what extent they are known and used in medical, epidemiological and economic research.

Estimating the Critical Parameter in Almost Stochastic Dominance from Insurance Deductibles

2020 ◽  
Author(s):  
Yi-Chieh Huang ◽  
Kamhon Kan ◽  
Larry Y. Tzeng ◽  
Kili C. Wang
Keyword(s):  
Real World ◽  
Upper Bound ◽  
Stochastic Dominance ◽  
Critical Parameter ◽  
Experimental Studies ◽  
Decision Makers ◽  
The Other ◽  
Real World Data ◽  
Upper Bound Estimate ◽  
Almost Stochastic Dominance

Knowing how small a violation of stochastic dominance rules would be accepted by most individuals is a prerequisite to applying almost stochastic dominance criteria. Unlike previous laboratory-experimental studies, this paper estimates an acceptable violation of stochastic dominance rules with 939,690 real world data observations on a choice of deductibles in automobile theft insurance. We find that, for all policyholders in the sample who optimally chose a low deductible, the upper bound estimate of the acceptable violation ratio is 0.0014, which is close to zero. On the other hand, considering that most decision makers, such as 99% (95%) of the policyholders in the sample, optimally chose the low deductible, the upper bound estimate of the acceptable violation ratio is 0.0405 (0.0732). Our results provide reference values for the acceptable violation ratio for applying almost stochastic dominance rules. This paper was accepted by Manel Baucells, decision analysis.

scholarly journals A scoping review of core outcome sets and their ‘mapping’ onto real-world data using prostate cancer as a case study

2020 ◽  
Vol 20 (1) ◽  
Author(s):  
Michela Meregaglia ◽  
Oriana Ciani ◽  
Helen Banks ◽  
Maximilian Salcher-Konrad ◽  
Caroline Carney ◽  
...  
Keyword(s):  
Prostate Cancer ◽  
Scoping Review ◽  
Real World ◽  
Core Outcome ◽  
Real World Data ◽  
Core Outcome Sets ◽  
World Data
Sign in / Sign up

Export Citation Format

Share Document