ABSTRACT
The previous eight chapters have been in preparation of this chapter. One of the main goals of this book is to show the reader how the advancement of lung CT AI over the past 45 years has made it possible to provide lung CT AI technologies in the current clinical medical imaging environment. This chapter will first describe the current information technologies (IT) that are in use in modern healthcare hospitals and clinics to support the acquisition, storage, and distribution of medical imaging studies including x-ray CT studies of the thorax, and how imaging IT interacts with the larger healthcare IT that supports the patient's electronic medical record (EMR). Then we will discuss the lung CT AI technology currently available within the imaging IT ecosystem to automatically assess lung CT images for quantitative metrics of lung disease. Specifically, we will discuss VIDA Insights v3.0 Density/tMPR module and Texture/Subpleural View module that use reactive machine AI and limited-memory AI methods to analyze each lung and lobe of a chest CT scan for the following quantitative CT metrics: volume in liters, LAA−950 metric for emphysema, HAA−700 to −250 metric for COVID-19 pneumonia and ILD, and the texture patterns that make up the HAA−700 to −250 that include ground-glass/reticular opacities, consolidation, and honeycombing. The importance of responsible AI will be discussed along with guidelines to achieve responsible lung CT AI.
ABSTRACT
In this chapter, we discuss the use of limited memory AI computer programs in lung CT AI for the assessment of normal and diseased lung tissue. Limited memory AI methods were discussed in Chapter 4 in the assessment of pulmonary nodules. Limited memory AI programs go beyond the reactive machine AI methods that were discussed in Chapters 5 and 6. Lung CT AI methods that use limited memory AI can detect and assess lung disease from COPD, ILD, and COVID-19 pneumonia in ways that reactive machine AI methods cannot. Lung CT AI using limited memory AI can detect and assess different lung CT texture patterns in patients with ILD including ground-glass opacities, reticulations, and honeycombing. Limited memory AI methods applied to lung CT AI can detect and distinguish different texture patterns of emphysema in COPD patients better than visual CT methods. Limited memory lung CT AI can detect, assess, and distinguish COVID-19 viral pneumonia from other forms of pneumonia.
ABSTRACT
This chapter will build on the lung CT AI approaches to lung nodules that were described in Chapter 4. In this chapter, we will describe the basic structure of the normal human lung and how emphysema from COPD decreases lung density that can be detected and assessed using reactive machine lung CT AI methods. Pulmonary fibrosis from ILD increases lung density as does acute viral pneumonia, such as COVID-19, and this increase in lung density can be detected and assessed using reactive machine-type lung CT AI methods. Four main steps in implementing lung CT AI to assess normal and abnormal structure in the human lung will be discussed;these include high-quality chest CT, segmenting the lung, extracting quantitative CT (QCT) features from the lung, using the QCT features to detect and assess diffuse lung disease. This chapter focuses on structural abnormalities of the lung that can be determined using a single chest CT scan of both lungs that is obtained at total lung capacity (TLC).
ABSTRACT
BACKGROUND: While international guidelines recommend medication reviews as part of the management of multimorbidity, evidence on how to implement reviews in practice in primary care is lacking. The MyComrade (MultimorbiditY Collaborative Medication Review And Decision Making) intervention is an evidence-based, theoretically informed novel intervention which aims to support the conduct of medication reviews for patients with multimorbidity in primary care. AIM: The pilot study aimed to assess the feasibility of a definitive trial of the MyComrade intervention across two healthcare systems (Republic of Ireland (ROI) and Northern Ireland (NI)). DESIGN: A pilot cluster-randomised controlled trial was conducted (clustered at general practice level), using specific progression criteria and a process evaluation framework. SETTING: General practices in the ROI and NI. PARTICIPANTS: Eligible practices were those in defined geographical areas who had GP's and Practice Based Pharmacists (PBP's) (in NI) willing to conduct medication reviews. Eligible patients were those aged 18 years and over, with multi morbidity and on ten or more medications. INTERVENTION: The MyComrade intervention is an evidence-based, theoretically informed novel intervention which aims to support the conduct of medication reviews for patients with multimorbidity in primary care, using a planned collaborative approach guided by an agreed checklist, within a specified timeframe. OUTCOME MEASURES: Feasibility outcomes, using pre-determined progression criteria, assessed practice and patient recruitment and retention and intervention acceptability and fidelity. Anonymised patient-related quantitative data, from practice medical records and patient questionnaires were collected at baseline, 4 and 8 months, to inform potential outcome measures for a definitive trial. These included (i) practice outcomes-completion of medication reviews; (ii) patient outcomes-treatment burden and quality of life; (iii) prescribing outcomes-number and changes of prescribed medications and incidents of potentially inappropriate prescribing; and (iv) economic cost analysis. The framework Decision-making after Pilot and feasibility Trials (ADePT) in conjunction with a priori progression criteria and process evaluation was used to guide the collection and analysis of quantitative and qualitative data. RESULTS: The recruitment of practices (n = 15) and patients (n = 121, mean age 73 years and 51% female), representing 94% and 38% of a priori targets respectively, was more complex and took longer than anticipated; impacted by the global COVID-19 pandemic. Retention rates of 100% of practices and 85% of patients were achieved. Both practice staff and patients found the intervention acceptable and reported strong fidelity to the My Comrade intervention components. Some practice staff highlighted concerns such as poor communication of the reviews to patients, dissatisfaction regarding incentivisation and in ROI the sustainability of two GPs collaboratively conducting the medication reviews. Assessing outcomes from the collected data was found feasible and appropriate for a definitive trial. Two progression criteria met the 'Go' criterion (practice and patient retention), two met the 'Amend' criterion (practice recruitment and intervention implementation) and one indicated a 'Stop - unless changes possible' (patient recruitment). CONCLUSION: The MyComrade intervention was found to be feasible to conduct within two different healthcare systems. Recruitment of participants requires significant time and effort given the nature of this population and the pairing of GP and pharmacist may be more sustainable to implement in routine practice. TRIAL REGISTRATION: Registry: ISRCTN, ISRCTN80017020 ; date of confirmation 4/11/2019; retrospectively registered.
ABSTRACT
BACKGROUND: The D1 Now intervention is designed to improve outcomes in young adults living with type 1 diabetes. It consists of three components: an agenda-setting tool, an interactive messaging system and a support worker. The aim of the D1 Now pilot cluster randomised controlled trial (RCT) was to gather and analyse acceptability and feasibility data to allow (1) further refinement of the D1 Now intervention, and (2) determination of the feasibility of evaluating the D1 Now intervention in a future definitive RCT. METHODS: A pilot cluster RCT with two intervention arms and a control arm was conducted over 12 months. Quantitative data collection was based on a core outcome set and took place at baseline and 12 months. Semi-structured interviews with participants took place at 6, 9 and 12 months. Fidelity and health economic costings were also assessed. RESULTS: Four diabetes centres and 57 young adults living with type 1 diabetes took part. 50% of eligible young adults were recruited and total loss to follow-up was 12%. Fidelity, as measured on a study delivery checklist, was good but there were three minor processes that were not delivered as intended in the protocol. Overall, the qualitative data demonstrated that the intervention was considered acceptable and feasible, though this differed across intervention components. The agenda-setting tool and support worker intervention components were acceptable to both young adults and staff, but views on the interactive messaging system were mixed. CONCLUSIONS: Some modifications are required to the D1 Now intervention components and research processes but with these in place progression to a definitive RCT is considered feasible. TRIAL REGISTRATION: ISRCTN (ref: ISRCTN74114336 ).