1
|
Lenskjold A, Brejnebøl MW, Rose MH, Gudbergsen H, Chaudhari A, Troelsen A, Moller A, Nybing JU, Boesen M. Artificial intelligence tools trained on human-labeled data reflect human biases: a case study in a large clinical consecutive knee osteoarthritis cohort. Sci Rep 2024; 14:26782. [PMID: 39500908 PMCID: PMC11538298 DOI: 10.1038/s41598-024-75752-z] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/28/2024] [Accepted: 10/08/2024] [Indexed: 11/08/2024] Open
Abstract
Humans have been shown to have biases when reading medical images, raising questions about whether humans are uniform in their disease gradings. Artificial intelligence (AI) tools trained on human-labeled data may have inherent human non-uniformity. In this study, we used a radiographic knee osteoarthritis external validation dataset of 50 patients and a six-year retrospective consecutive clinical cohort of 8,273 patients. An FDA-approved and CE-marked AI tool was tested for potential non-uniformity in Kellgren-Lawrence grades between the right and left sides of the images. We flipped the images horizontally so that a left knee looked like a right knee and vice versa. According to human review, the AI tool showed non-uniformity with 20-22% disagreements on the external validation dataset and 13.6% on the cohort. However, we found no evidence of a significant difference in the accuracy compared to senior radiologists on the external validation dataset, or age bias or sex bias on the cohort. AI non-uniformity can boost the evaluated performance against humans, but image areas with inferior performance should be investigated.
Collapse
Affiliation(s)
- Anders Lenskjold
- Department of Radiology, Copenhagen University Hospital Bispebjerg-Frederiksberg, Copenhagen, Denmark.
- Radiological Artificial Intelligence Testcenter, Copenhagen, Denmark.
- Department of Clinical Medicine, University of Copenhagen University, Copenhagen, Denmark.
| | - Mathias W Brejnebøl
- Department of Radiology, Copenhagen University Hospital Bispebjerg-Frederiksberg, Copenhagen, Denmark
- Radiological Artificial Intelligence Testcenter, Copenhagen, Denmark
- Department of Clinical Medicine, University of Copenhagen University, Copenhagen, Denmark
| | - Martin H Rose
- Center for Surgical Science, Zealand University Hospital, Køge, Denmark
| | - Henrik Gudbergsen
- The Parker Institute, University of Copenhagen, Copenhagen, Denmark
- Department of Public Health, Center for General Practice, University of Copenhagen, Copenhagen, Denmark
| | - Akshay Chaudhari
- Department of Radiology, Stanford University, Stanford, CA, USA
- Department of Biomedical Data Science, Stanford University, Stanford, CA, USA
| | - Anders Troelsen
- Department of Clinical Medicine, University of Copenhagen University, Copenhagen, Denmark
- Department of Orthopaedic Surgery, Copenhagen University Hospital Hvidovre & CAG, ROAD - Research OsteoArthritis, Hvidovre, Denmark
| | - Anne Moller
- Department of Public Health, Center for General Practice, University of Copenhagen, Copenhagen, Denmark
- Primary Health Care Research Unit, Region Zealand, Denmark
| | - Janus U Nybing
- Department of Radiology, Copenhagen University Hospital Bispebjerg-Frederiksberg, Copenhagen, Denmark
- Radiological Artificial Intelligence Testcenter, Copenhagen, Denmark
| | - Mikael Boesen
- Department of Radiology, Copenhagen University Hospital Bispebjerg-Frederiksberg, Copenhagen, Denmark
- Radiological Artificial Intelligence Testcenter, Copenhagen, Denmark
- Department of Clinical Medicine, University of Copenhagen University, Copenhagen, Denmark
| |
Collapse
|
2
|
Including mental health care in a model of European health system. Epidemiol Psychiatr Sci 2023; 32:e12. [PMID: 36803918 PMCID: PMC9971856 DOI: 10.1017/s2045796023000057] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Indexed: 02/22/2023] Open
Abstract
The management of a health system is a matter of economics and business administration because of the costs induced by goods and services delivered. Economics teaches us that the positive effects induced by competition in free markets cannot be expected in health care, which is a classic example of market failure from both demand and supply sides. The most sensible key concepts to refer for managing a health system are funding and provision. While the logical solution for the first variable is universal coverage through general taxation, the second one requires a deeper understanding. Integrated care is the modern approach that better supports the choice in favour of the public sector also for service provision. A major threat against this approach is dual practice legally allowed for health professionals, which inevitably raises financial conflicts of interest. An exclusive contract of employment for civil servants should be the sine qua non for providing public services effectively and efficiently. Integrated care is particularly important for long-term chronic illnesses associated with high levels of disability, such as neurodegenerative diseases and mental disorders, where the mix of health and social services needed can be very complex. Nowadays the growing number of community-dwelling patients with multiple physical and mental health needs is the major challenge for the European health systems. This happens also in public health systems, which should provide universal health coverage in principle, and the case of mental disorders is striking. In the light of this theoretical exercise, we strongly believe that a public National Health and Social Service should be the most indicated model for both funding and providing health and social care in modern societies. The big challenge of the common model of European health system here envisaged would be to limit the negative influences of politics and bureaucracy.
Collapse
|
3
|
Rosół I, Ciesielka J, Matlakiewicz M, Grześków M, Cebula M, Gruszczyńska K, Winder M. The Assessment of the Rationale for Urgent Head CT-Comparative Analysis of Referrals and Results of Examinations without and with Contrast Enhancement. MEDICINA (KAUNAS, LITHUANIA) 2022; 58:medicina58101468. [PMID: 36295628 PMCID: PMC9610557 DOI: 10.3390/medicina58101468] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Download PDF] [Figures] [Subscribe] [Scholar Register] [Received: 08/29/2022] [Revised: 10/03/2022] [Accepted: 10/14/2022] [Indexed: 11/06/2022]
Abstract
The study analyzes the correlation between the indications and results of head CT examinations in search of evidence of the excessive use of this diagnostic method. In total, 1160 referrals for urgent head CT were analyzed retrospectively, including the following parameters: patients’ sex and age, type of scan (C−, C+, angio-CT), description of symptoms and presence of diagnostic target. Pathologies identified by the radiologist were assigned to four classes, regarding the severity of diagnosed conditions. The analysis of the CT results has shown that over half (55.22%) of the examinations revealed no deviations or showed chronic, asymptomatic lesions. As many as 73.71% referrals constituted group 0 in terms of the lack of a diagnostic target of a specific pathology. The presence of specific clinical targeting in a referral correlated significantly with a higher frequency of acute diagnosis. Contrast-enhanced follow-up examinations allowed the unequivocal classification of patients into extreme classes (I or IV) and accurate identification of patients requiring urgent or chronic treatment. Excessive use of diagnostic imaging is harmful, not only to patients, who often are unnecessarily exposed to radiation, but also to the quality of healthcare, since it increases the costs and radiologists’ workload.
Collapse
Affiliation(s)
- Izabela Rosół
- Students’ Scientific Society, Department of Radiology and Nuclear Medicine, Medical University of Silesia, ul. Medyków 14, 40-752 Katowice, Poland
| | - Jakub Ciesielka
- Students’ Scientific Society, Department of Radiology and Nuclear Medicine, Medical University of Silesia, ul. Medyków 14, 40-752 Katowice, Poland
| | - Magdalena Matlakiewicz
- Students’ Scientific Society, Department of Radiology and Nuclear Medicine, Medical University of Silesia, ul. Medyków 14, 40-752 Katowice, Poland
| | - Michał Grześków
- Students’ Scientific Society, Department of Radiology and Nuclear Medicine, Medical University of Silesia, ul. Medyków 14, 40-752 Katowice, Poland
| | - Maciej Cebula
- Department of Radiology and Nuclear Medicine, Medical University of Silesia, ul. Medyków 14, 40-752 Katowice, Poland
| | - Katarzyna Gruszczyńska
- Department of Radiology and Nuclear Medicine, Medical University of Silesia, ul. Medyków 14, 40-752 Katowice, Poland
| | - Mateusz Winder
- Department of Radiology and Nuclear Medicine, Medical University of Silesia, ul. Medyków 14, 40-752 Katowice, Poland
- Correspondence: ; Tel.: +48-32-789-47-51
| |
Collapse
|
4
|
Miziara ID, Miziara CSMG. Medical errors, medical negligence and defensive medicine: A narrative review. Clinics (Sao Paulo) 2022; 77:100053. [PMID: 35640458 PMCID: PMC9160317 DOI: 10.1016/j.clinsp.2022.100053] [Citation(s) in RCA: 5] [Impact Index Per Article: 1.7] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Download PDF] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 03/22/2022] [Accepted: 05/17/2022] [Indexed: 11/03/2022] Open
Abstract
Error in medicine and medical liability has a long history dating back to Antiquity. During the 19th Century, most lawsuits related to errors in treating surgical problems were settled. However, in the first half of the 20th Century, lawsuits claimed that mistakes were related to the doctor's action: the doctor made something wrong (errors of commission). In Brazil, medical error is defined as inappropriate conduct, including negligence and recklessness, that causes harm to the patient. The physician's fear of being suited is the reason for some practice named defensive Medicine (D.M.), defined as ordering unnecessary tests and procedures or avoiding treatments for patients considered at high-risk. Thus, this narrative review aims to analyze and describe the relationship between medical errors, medical negligence, and the practice of D.M. So, the authors propose procedures and attitudes to avoid medical errors and the approach of D.M.: a national focus to create leadership and research tools to enhance the knowledge base about patient safety; a reporting system that would help to identify and learn from errors; the use of a computer-based protocol reminder; some technological devices to help the medical practice (electronic prescribing and information technology systems); creating risk management programs in hospitals. Therefore, the authors conclude that the most critical attitude to avoid medical liability is a good and ethical medical practice with the proper use of technology, based on knowledge of scientific evidence and ethical principles of medicine - for the benefit of patients.
Collapse
Affiliation(s)
- Ivan Dieb Miziara
- Department of Legal Medicine, Ethics and Occupational Health, Faculdade de Medicina da Universidade de São Paulo, São Paulo, SP, Brazil.
| | | |
Collapse
|
5
|
Garattini L, Badinella Martini M, Mannucci PM. Integrated care: easy in theory, harder in practice? Intern Emerg Med 2022; 17:3-6. [PMID: 34491514 DOI: 10.1007/s11739-021-02830-9] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 07/08/2021] [Accepted: 08/14/2021] [Indexed: 10/20/2022]
Abstract
Integrated care (IC) is a term now commonly adopted across the world, which implies a positive attitude towards addressing fragmentation of service provision inside health systems. While the principles of IC are simple, their implementation is more controversial. The ever growing number of IC definitions is related to the increasing domains of applications, which reflect the increasing demand induced by ageing multi-morbid patients. An exhaustive definition of IC should now enclose the coordination of health and social services useful to deliver seamless care across organizational boundaries. The current debate on IC is largely fueled by the modern mismatch between the growing burden of health needs for chronic conditions from the demand side and the design of health systems still largely centered on acute care from the supply side. The major reasons of persisting IC weakness in Western European nations stem from arguable choices of health policy taken in a quite recent past. The political creed in 'market competition' is likely to be the most emblematic. All initiatives encouraging healthcare providers to compete with each other are likely to discourage IC. Another historically rooted reason of IC weakness is the occupational status of European general practitioners (GPs). While single large-scale organizations have become a pressing priority for a modern primary care, most GPs are still selfemployed professionals working in their own cabinets. It is time to reconsider the anachronistic status of GPs so as to enhance IC in the future.
Collapse
Affiliation(s)
- Livio Garattini
- Institute for Pharmacological Research Mario Negri IRCCS, Ranica, BG, Italy.
| | | | | |
Collapse
|
6
|
Varley E, Varma S. Introduction: medicine's shadowside: revisiting clinical iatrogenesis. Anthropol Med 2021; 28:141-155. [PMID: 34355978 DOI: 10.1080/13648470.2021.1937514] [Citation(s) in RCA: 5] [Impact Index Per Article: 1.3] [Reference Citation Analysis] [Abstract] [Key Words] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 10/20/2022]
Abstract
Drawing on the work of Ivan Illich, our special issue reanimates iatrogenesis as a vital concept for the social sciences of medicine. It calls for medicine to expand its engagement of the injustices that unfold from clinical processes, practices, and protocols into patient lifeworlds and subjectivities beyond the clinic. The capacious view of iatrogenesis revealed by this special issue collection affords fuller and more heterogeneous insights on iatrogenesis that does not limit it to medical explanations alone, nor locate harm in singular points in time. These papers attend to iatrogenesis' immediate and lingering presences in socialities and structures within and beyond medicine, and the ways it reflects or reproduces the racism, sexism, and ableism built into medical logics.
Collapse
Affiliation(s)
- Emma Varley
- Anthropology, Brandon University, Brandon, Manitoba, Canada
| | - Saiba Varma
- Anthropology, University of California, San Diego, La Jolla, CA, USA
| |
Collapse
|