Objectives for algorithmic decision-making systems in childhood asthma: Perspectives of children, parents, and physicians

Abstract

Objectives: To identify with children, parents and physicians the objectives to be used as parameters for algorithmic decision-making systems (ADMSs) adapting treatments in childhood asthma.

Methods: We first conducted a qualitative study based on semi-structured interviews to explore the objectives that children aged 8-17 years, their parents, and their physicians seek to achieve when taking/giving/prescribing a treatment for asthma. Following the grounded theory approach, each interview was independently coded by two researchers; reconciled codes were used to assess code frequency, categories were defined, and the main objectives identified. We then conducted a quantitative study based on questionnaires using these objectives to determine how children/parents/physicians ranked these objectives and whether their responses were aligned.

Results: We interviewed 71 participants (31 children, 30 parents and 10 physicians) in the qualitative study and identified seven objectives associated with treatment uptake and five objectives associated with treatment modalities. We included 291 participants (137 children, 137 parents, and 17 physicians) in the quantitative study. We found little correlation between child, parent, and physician scores for each of the objectives. Each child's asthma history influenced the choice of scores assigned to each objective by the child, parents, and physician.

Conclusion: The identified objectives are quantifiable and relevant to the management of asthma in the short and long term. They can therefore be incorporated as parameters for future ADMS. Shared decision-making seems essential to achieve consensus among children, parents, and physicians when choosing the weight to assign to each of these objectives.

Keywords: Decision-making: computer-assisted; algorithms; asthma/drug therapy; child; shared decision-making.

Conflict of interest statement

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

© The Author(s) 2024.

Figures

Figure 1.

Development of the codebook.

Figure 2.

Cluster of words obtained from the automated textual analysis. Words association automatically identified by the Software IRaMuTeQ. The size of each word is proportionate to the frequency of its use during the interviews.

Figure 3.

Flow chart of the surveys analyzed in the quantitative study.

Figure 4.

Comparison of scores assigned to each treatment objective by children, parents, and physicians. Violins plots are presented: the length of the colored area illustrates the distribution of scores, with wider sections indicating a greater proportion of participants giving that score. In each violin diagram, the box plot with median and interquartile values is presented. The child's responses are shown in red, those of the parents in green and those of the doctor in blue. NS: not statistically significant; *p < .05; **p < .01.

Figure 5.

Correlations of scores assigned overall and to each objective by children, their parents, and their physicians.

Figure 6.

Factors associated with the choice of each objective for children (in red), parents (in green), and physicians (in blue). ACT: asthma control test.

Figure 7.

Factors associated with the choice of each treatment modality objective for children (in red), parents (in green), and physicians (in blue).