Systematic review of deep learning image analyses for the diagnosis and monitoring of skin disease.

Skin diseases affect one-third of the global population, posing a major healthcare burden. Deep learning may optimise healthcare workflows through processing skin images via neural networks to make predictions. A focus of deep learning research is skin lesion triage to detect cancer, but this may not translate to the wider scope of >2000 other skin diseases. We searched for studies applying deep learning to skin images, excluding benign/malignant lesions (1/1/2000-23/6/2022, PROSPERO CRD42022309935). The primary outcome was accuracy of deep learning algorithms in disease diagnosis or severity assessment. We modified QUADAS-2 for quality assessment. Of 13,857 references identified, 64 were included. The most studied diseases were acne, psoriasis, eczema, rosacea, vitiligo, urticaria. Deep learning algorithms had high specificity and variable sensitivity in diagnosing these conditions. Accuracy of algorithms in diagnosing acne (median 94%, IQR 86-98; n = 11), rosacea (94%, 90-97; n = 4), eczema (93%, 90-99; n = 9) and psoriasis (89%, 78-92; n = 8) was high. Accuracy for grading severity was highest for psoriasis (range 93-100%, n = 2), eczema (88%, n = 1), and acne (67-86%, n = 4). However, 59 (92%) studies had high risk-of-bias judgements and 62 (97%) had high-level applicability concerns. Only 12 (19%) reported participant ethnicity/skin type. Twenty-four (37.5%) evaluated the algorithm in an independent dataset, clinical setting or prospectively. These data indicate potential of deep learning image analysis in diagnosing and monitoring common skin diseases. Current research has important methodological/reporting limitations. Real-world, prospectively-acquired image datasets with external validation/testing will advance deep learning beyond the current experimental phase towards clinically-useful tools to mitigate rising health and cost impacts of skin disease.

Innovative regenerative medicines in the EU: a better future in evidence?

BACKGROUND: Despite a steady stream of headlines suggesting they will transform the future of healthcare, high-tech regenerative medicines have, to date, been quite inaccessible to patients, with only eight having been granted an EU marketing licence in the last 7 years. Here, we outline some of the historical reasons for this paucity of licensed innovative regenerative medicines. We discuss the challenges to be overcome to expedite the development of this complex and rapidly changing area of medicine, together with possible reasons to be more optimistic for the future. DISCUSSION: Several factors have contributed to the scarcity of cutting-edge regenerative medicines in clinical practice. These include the great expense and difficulties involved in planning how individual therapies will be developed, manufactured to commercial levels and ultimately successfully delivered to patients. Specific challenges also exist when evaluating the safety, efficacy and cost-effectiveness of these therapies. Furthermore, many treatments are used without a licence from the European Medicines Agency, under "Hospital Exemption" from the EC legislation. For products which are licensed, alternative financing approaches by healthcare providers may be needed, since many therapies will have significant up-front costs but uncertain benefits and harms in the long-term. However, increasing political interest and more flexible mechanisms for licensing and financing of therapies are now evident; these could be key to the future growth and development of regenerative medicine in clinical practice. CONCLUSIONS: Recent developments in regulatory processes, coupled with increasing political interest, may offer some hope for improvements to the long and often difficult routes from laboratory to marketplace for leading-edge cell or tissue therapies. Collaboration between publicly-funded researchers and the pharmaceutical industry could be key to the future development of regenerative medicine in clinical practice; such collaborations might also offer a possible antidote to the innovation crisis in the pharmaceutical industry.

Randomised trials comparing different healthcare settings: an exploratory review of the impact of pre-trial preferences on participation, and discussion of other methodological challenges.

BACKGROUND: We recently published a systematic review of different healthcare settings (such as outpatient, community or home) for administering intravenous chemotherapy, and concluded that performing conventionally designed randomised trials was difficult. The main problems were achieving adequate trial accrual rates and recruiting a study population which adequately represented the target population of interest. These issues stemmed from the fact that potential participants may have had pre-trial perceptions about the trial settings they may be allocated; such preferences will sometimes be strong enough for patients to decline an invitation to participate in a trial. A patient preference trial design (in which patients can choose, or be randomised to, an intervention) may have obviated these recruitment issues, although none of the trials used such a design. METHODS: In order to gain a better understanding of the broader prevalence and extent of these preference issues (and any other methodological challenges), we undertook an exploratory review of settings trials in any area of healthcare treatment research. We searched The Cochrane Library and Google Scholar and used snowballing methods to identify trials comparing different healthcare settings. RESULTS: Trial accrual was affected by patient preferences for a setting in 15 of the 16 identified studies; birth setting trials were the most markedly affected, with between 68 % and 85 % of eligible women declining to participate specifically because of preference for a particular healthcare setting. Recruitment into substance abuse and chemotherapy setting studies was also notably affected by preferences. Only four trials used a preference design: the proportion of eligible patients choosing to participate via a preference group ranged from between 33 % and 67 %. CONCLUSIONS: In trials of healthcare settings, accrual may be seriously affected by patient preferences. The use of trial designs which incorporate a preference component should therefore strongly be considered. When designing such trials, investigators should consider settings to be complex interventions, which are likely to have linked components which may be difficult to control for. Careful thought is also needed regarding the choice of comparator settings and the most appropriate outcome measures to be used.

Assessing baseline imbalance in randomised trials: implications for the Cochrane risk of bias tool.

A key component of the Cochrane Collaboration's risk of bias tool for critically evaluating randomised trials is the consideration of whether baseline characteristics of the treatment groups being compared are systematically different. Considered under the domain of 'selection bias', this is currently evaluated by looking at the methods of randomisation and specifically at the generation of the randomised allocation sequence and the concealment of this sequence during the process of randomisation. Assessment of the actual similarity of baseline variables across groups in demographic and clinical characteristics is seldom performed. Even when performed, the link with selection bias is sometimes not considered. Methods of randomisation and allocation concealment are often poorly reported in published trials, yet baseline data tables are presented in a large majority of trial reports. In this article, we propose that assessment of trial baseline data should form a key and prominent part of selection bias judgements when using the risk of bias tool. We outline the possible benefits from using this approach, including reduced uncertainty in systematic review conclusions, reduced risk of chance findings being ascribed to treatment effects and better use of available evidence by a more considered approach to evaluating studies using imperfect randomisation and allocation methods.

The use and reporting of WOMAC in the assessment of the benefit of physical therapies for the pain of osteoarthritis of the knee: findings from a systematic review of clinical trials.

OBJECTIVE: For the purposes of meta-analysis and network meta-analysis, the use of standard outcome measures is ideal. In OA research, the WOMAC was developed as an OA-specific measure of disability. It includes a pain subscale. In 1994 a consensus meeting recommended the use of WOMAC as a primary measure of efficacy in OA. In the context of a review of the efficacy of physical interventions for the relief of the pain of OA of the knee, we investigated the use of WOMAC. METHODS: A systematic review (December 2009-January 2010) identified trials that used the WOMAC outcome. These were investigated for correct use and clear reporting of the WOMAC pain subscale and the WOMAC index. RESULTS: The WOMAC pain subscale was used in 45% of the 134 trials. Reporting of the exact method of administering the WOMAC pain subscale was poor in many cases: in 53% of trials the reporting of the type of WOMAC scale used was inadequate; the score range was reported ambiguously in 38% of trials, with a further 10% being completely unclear. Similar less than optimal reporting of the WOMAC index was found. CONCLUSION: Poor reporting of both the WOMAC pain subscale and the WOMAC index resulted in significant uncertainty in the interpretation of the results of individual trials and limited their contribution to evidence synthesis. Improved adherence with the standard use of the WOMAC scoring system, with clear reporting of it in trials of OA of the knee should be encouraged.