One of the main challenges in testing artificial intelligence (AI) enabled cyber physicalsystems (CPS) such as autonomous driving systems and internet­-of-­things (IoT) medical
devices is the presence of machine learning components, for which formal properties are
difficult to establish. In addition, operational components interaction circumstances, inclusion of human­-in-­the-­loop, and environmental changes result in a myriad of safety concerns
all of which may not only be comprehensibly tested before deployment but also may not
even have been detected during design and testing phase. This dissertation identifies major challenges of safety verification of AI­-enabled safety critical systems and addresses the
safety problem by proposing an operational safety verification technique which relies on
solving the following subproblems:
1. Given Input/Output operational traces collected from sensors/actuators, automatically
learn a hybrid automata (HA) representation of the AI-­enabled CPS.
2. Given the learned HA, evaluate the operational safety of AI­-enabled CPS in the field.
This dissertation presents novel approaches for learning hybrid automata model from time
series traces collected from the operation of the AI­-enabled CPS in the real world for linear
and non­linear CPS. The learned model allows operational safety to be stringently evaluated
by comparing the learned HA model against a reference specifications model of the system.
The proposed techniques are evaluated on the artificial pancreas control system

Background: Bacterial colonization of the respiratory tract is commonly described and usually thought to be of no clinical significance. The aim of this study was to examine the presence and significance of bacteria and viruses in the upper respiratory tract of healthcare workers (HCWs), and association with respiratory symptoms.

Methods: A prospective cohort study was conducted in China and 223 HCWs were recruited from fever clinics and respiratory, paediatric, emergency/Intensive medication wards. Participants were followed over 4 weeks (7th May 2015 to 4th June 2015) for development of clinical respiratory illness (CRI). Nasopharyngeal swabs were obtained at baseline and at the end of the study. The primary endpoints were laboratory-confirmed bacterial colonization and viral respiratory infection. Rates of the following infections in symptomatic and asymptomatic participants were compared at the start or end of the study; 1) all bacterial/viral infections, 2) bacterial infection and bacterial-viral co-infections, excluding virus only infections, and 3) only bacterial infections.

Results: Bacterial colonization was identified in 88% (196/223) of participants at the start or end of the study. Among these participants, 66% (148/223) had only bacterial colonization while 22% (48/223) had co-infection with a virus. Bacteria were isolated from 170 (76.2%) participants at baseline and 127 (57%) participants at the end of the study. Laboratory confirmed viral infections were identified in 53 (23.8%) participants - 35 (15.7%) at the baseline and 20 (9.0%) at the end of the study. CRI symptoms were recorded in 12 participants (4.5%) and all had a positive bacterium isolation at baseline (n = 11) or end of the study (n = 1). Among asymptomatic participants, 187 (87%) had bacterial colonization or bacterial/viral co-infection at baseline or end of the study. Viruses were also isolated from 5 (2.4%) asymptomatic cases. Rates of all infection outcomes were higher in symptomatic participants, however differences were not statistically significant.

Conclusion: We isolated high rates of bacteria and viruses in the upper respiratory tract of hospital HCWs, which may reflect greater exposure to respiratory infections in the hospital. Although respiratory infections are mostly symptomatic, the association between bacterial colonization and symptomatic illness is not clear. In the healthcare setting, HCWs may acquire and transmit infection to patients and other HCWs around them. Larger studies are required to explore ongoing occupational risk of respiratory infection in hospitals HCWs.

Rationale: Medical masks are commonly used by sick individuals with influenza-like illness (ILI) to prevent spread of infections to others, but clinical efficacy data are absent.

Objective: Determine whether medical mask use by sick individuals with ILI protects well contacts from related respiratory infections.

Setting: 6 major hospitals in 2 districts of Beijing, China.

Design: Cluster randomised controlled trial.

Participants: 245 index cases with ILI.

Intervention: Index cases with ILI were randomly allocated to medical mask (n=123) and control arms (n=122). Since 43 index cases in the control arm also used a mask during the study period, an as-treated post hoc analysis was performed by comparing outcomes among household members of index cases who used a mask (mask group) with household members of index cases who did not use a mask (no-mask group).

Main Outcome Measure: Primary outcomes measured in household members were clinical respiratory illness, ILI and laboratory-confirmed viral respiratory infection.

Results: In an intention-to-treat analysis, rates of clinical respiratory illness (relative risk (RR) 0.61, 95% CI 0.18 to 2.13), ILI (RR 0.32, 95% CI 0.03 to 3.13) and laboratory-confirmed viral infections (RR 0.97, 95% CI 0.06 to 15.54) were consistently lower in the mask arm compared with control, although not statistically significant. A post hoc comparison between the mask versus no-mask groups showed a protective effect against clinical respiratory illness, but not against ILI and laboratory-confirmed viral respiratory infections.

Conclusions: The study indicates a potential benefit of medical masks for source control, but is limited by small sample size and low secondary attack rates. Larger trials are needed to confirm efficacy of medical masks as source control.

Widespread adoption of smartphone based Mobile Medical Apps (MMAs) is opening new avenues for innovation, bringing MMAs to the forefront of low cost healthcare delivery. These apps often control human physiology and work on sensitive data. Thus it is necessary to have evidences of their trustworthiness i.e. maintaining privacy of health data, long term operation of wearable sensors and ensuring no harm to the user before actual marketing. Traditionally, clinical studies are used to validate the trustworthiness of medical systems. However, they can take long time and could potentially harm the user. Such evidences can be generated using simulations and mathematical analysis. These methods involve estimating the MMA interactions with human physiology. However, the nonlinear nature of human physiology makes the estimation challenging.

This research analyzes and develops MMA software while considering its interactions with human physiology to assure trustworthiness. A novel app development methodology is used to objectively evaluate trustworthiness of a MMA by generating evidences using automatic techniques. It involves developing the Health-Dev β tool to generate a) evidences of trustworthiness of MMAs and b) requirements assured code generation for vulnerable components of the MMA without hindering the app development process. In this method, all requests from MMAs pass through a trustworthy entity, Trustworthy Data Manager which checks if the app request satisfies the MMA requirements. This method is intended to expedite the design to marketing process of MMAs. The objectives of this research is to develop models, tools and theory for evidence generation and can be divided into the following themes:

• Sustainable design configuration estimation of MMAs: Developing an optimization framework which can generate sustainable and safe sensor configuration while considering interactions of the MMA with the environment.

• Evidence generation using simulation and formal methods: Developing models and tools to verify safety properties of the MMA design to ensure no harm to the human physiology.

• Automatic code generation for MMAs: Investigating methods for automatically

• Performance analysis of trustworthy data manager: Evaluating response time generating trustworthy software for vulnerable components of a MMA and evidences.performance of trustworthy data manager under interactions from non-MMA smartphone apps.