Charitable crowdfunding has revolutionized philanthropy and social support, providing a unique and accessible platform for individuals and organizations to raise funds for their causes. In the realm of medical crowdfunding, patients often turn to web-based platforms such as GoFundMe for financial support.

A critical element of a successful crowdfunding campaign is the selection of an effective campaign cover image. This image serves as a visual representation of the project and plays a key role in attracting potential donors [1-3]. It also helps to communicate the project’s core information and value proposition, allowing potential supporters to quickly understand the project’s goals and importance of the project [4,5].

In the current digital age, the campaign cover image serves as a first impression. Given people’s limited attention span, it has a significant impact on the outcome of the crowdfunding campaign [6,7]. Research has shown that individuals process and retain visual information better than textual data [8-12]. The importance of visual information has been demonstrated in advertising, marketing, technology crowdfunding, and other fields [13-16]. Researchers have investigated several factors that influence the effectiveness of crowdfunding campaigns, such as project descriptions, incentives, and engagement on social media platforms [17-21]. While there is growing public awareness of using web-based crowdfunding platforms to raise medical funds for patients with cancer, there is a lack of research in academia on how specific elements of a cancer crowdfunding project’s image affect its success rates.

In order to obtain a more profound understanding of the influence of images on the success of cancer crowdfunding campaigns, we performed a study of cancer-related crowdfunding campaigns on GoFundMe, which is the most extensive site for medical crowdfunding. Our hypotheses are shown in Textbox 1.

As shown in Figure 1, this section describes the flow of the research throughout the article in 5 parts.

We used a Python-based web-scraping program to collect data on all cancer-related crowdfunding projects hosted on the platform from January 1, 2019, to July 28, 2020. Ultimately, our efforts yielded data from a total of 307,886 projects available on the site. The dataset includes essential project details such as basic information, textual content, and images. In addition, the basic information includes several key variables that are critical for later analysis, including funding amount, number of donations, description length, title length, and duration. These elements are referred to as basic features.

Given that the United States is a major player in the global crowdfunding market, with a well-established environment and individuals often demonstrating a higher level of engagement and trust in crowdfunding efforts, we focused our initial screening on US-based projects with individual crowdfunding initiators. This refinement narrowed our dataset to 243,795 medical campaigns. We then transformed the dataset to facilitate analysis. This involved converting the fundraising amounts from textual to numerical values and calculating the time elapsed since the creation of each campaign using a predetermined base date.

By collecting data and identifying project features, we had gathered specific variables that could potentially affect cancer crowdfunding projects. These variables consist of project basic information, cancer types, image features in crowdfunding images, and the topic of image content.

Funding Percentage is a continuous variable, indicating the percentage of the target amount that the current project has raised. We used Funding Percentage to measure the outcome of a crowdfunding project.

In the process of selecting the independent variables, we considered a combination of interdisciplinary literature [22-27] that identifies individual characteristics such as the number of faces, gender, and facial expression. While these factors have been empirically supported in other domains, their effects have yet to be validated in the specific context of health care crowdfunding images. Furthermore, given the centrality of image content in the operational strategies of crowdfunding platforms such as Indiegogo, the visual presentation of different thematic content has been shown to significantly modulate crowdfunding outcomes [35-37]. Therefore, the inclusion of “topic” as an independent variable aims to analyze how it plays a unique role in the field of health care crowdfunding (Table 3).

In order to investigate the influence of image characteristics on crowdfunding results, a logistic regression model was used. In this study, regression models were built to predict crowdfunding outcomes in the presence of multicollinearity as determined by the variance inflation factor value. Additionally, the regression results were visualized as a word cloud to visually illustrate the content of different image themes and their effects by using Python’s WordCloud package.

The data sources for this study were information available on public platforms, and data collection did not involve direct participation of study participants. To protect privacy, all data were anonymized to ensure that individuals could not be identified from the results. Additionally, all data were stored on a secure server with access restricted to project team members, in accordance with a strict privacy policy.

Of the 243,795 images, 205,496 contained depictions of human figures, indicating a Face Count ≥1. The dataset was largely dominated by single-person images, accounting for 58.83% (120,893/205,496) of the total. Multiperson images included from 2 to 10 individuals. Regarding gender distribution, 42.31% (86,945/205,496) of images only contained male individuals, while images featuring only female individuals constituted 29.83% (61,230/205,496). The remaining 27.86% (57,251/205,496) featured a mix of male and female individuals. In total, 58.11% (119,413/205,496) of the total images depicted smiling faces. Moreover, 21.81% (44,818/205,496) of individuals wore hats.

To explore how the individuals’ content in the images affects the funding outcome, we built regression models for the attributes. We found the following variables to have a significant impact on funding outcomes. Table 4 presents the outcomes from the regression model.

Table 4 includes 4 types of individual variables that have a significant effect on funding outcomes.

The word cloud analysis from the previous section displays that images possessing diverse character traits have distinct scenarios and contents. Image content and individual traits exert significant influences on crowdfunding results, respectively. Table 8 illustrates the effects on crowdfunding results when both character traits and image content appear in an image.

The analysis revealed that for certain themes—namely food and cooking, city life, and medical and health—the interaction term Topic×Face Count showed a positive and significant coefficient (β values of 0.005, 0.0175, and 0.0118, and P values of .008, .007, and .005, respectively), indicating that an increase in the number of faces in these themed images correlates with a higher fundraising proportion.

Conversely, the Topic×Female Percent interaction term suggested that a higher percentage of female faces intensifies the negative impact when negative topics are present, particularly in rest and play and medical and health themes. However, an increase in Female Percent within the diverse locations theme positively influenced fundraising, as evidenced by a negative coefficient for Female Percent (β=–0.0201; P<.001) being offset by the positive interaction term Topic×Female Percent (β=0.0312; P<.001), resulting in a net positive effect.

When considering the presence of a smiley face in conjunction with the image’s topic, the rest and play theme correlated with a beneficial increase in funds raised (β=0.0152; P<.001). In contrast, themes such as city life (β=–0.0441; P=.002), clothing and color (β=–0.0919; P=.005), and diverse locations (β=–0.01; P=.006) were associated with a negative impact on fundraising.

To better understand the causal relationship between the key variables and crowdfunding success in the above study, we used counterfactual analysis, a widely used method for comparative investigation [53].

In the counterfactual analysis, the indicator of each significant variable is changed to zero, and the predicted values of the model are calculated, comparing the growth of the predicted values before and after the change. Table 9 and Table 10 show the results of the counterfactual analysis. The effect of the variable on the dependent variable as reflected in the results of the counterfactual analysis was consistent with the results of the regression model. For example, Face Count (P<.001) and If Smile (P<.001) had a positive causal effect on crowdfunding outcomes and Female Percent (P<.001) had a negative causal effect.

We found that certain characteristics of individuals in crowdfunding images can significantly influence the success or failure of crowdfunding activities. Specifically, the number of faces, age, gender, and facial expressions, such as smiling, were found to significantly affect the success of crowdfunding campaigns. First, previous studies have shown that the inclusion of faces in crowdfunding campaigns can significantly influence funders’ decisions [38,54]. Specifically, increasing the number of faces used leads to more positive effects. Similarly, our study found that obtaining more characters also leads to more donations. Second, our study suggests that cancer campaigns have better outcomes when a higher percentage of their images feature younger people. This finding is consistent with previous studies [37] suggesting that individuals tend to be more generous in their donations to children when viewing medical crowdfunding programs. Third, several studies [26-28] have demonstrated the positive impact of displaying smiles in crowdfunding projects, and our findings are consistent with this. Such images may contribute to a stronger first impression by increasing viewers’ empathy, which in turn may lead to better crowdfunding outcomes. Fourth, Elitzur and Solodoha [55] found a negative correlation between female leadership and funding success in rewards-based crowdfunding programs. In contrast, Figueroa-Armijos and Berns [56] reported that backers are more likely to support female-led projects than male-led projects. Elitzur and Solodoha [55] and Figueroa-Armijos and Berns [56] argue that the effect of gender on crowdfunding success varies according to the different patterns and stages of project development. However, our study shows that an increase in the proportion of female individuals in the image of medical crowdfunding cancer projects leads to a decrease in crowdfunding success.

For the thematic content of images, we found that images depicting the hair and beauty, medical and health, and food and cooking themes significantly impacted crowdfunding outcomes. The study suggests that these themes may help to awaken empathy and a sense of justice in viewers when viewing such images [57]. For example, beautiful images and photographic narratives related to patient health can evoke hope and empathy in viewers, ultimately encouraging donations [58,59]. However, certain image themes, such as city life, diverse locations, and rest and play, may negatively impact the outcome of a crowdfunding project. These themes are not relevant to medical crowdfunding content and may create additional challenges for readers to understand the primary message, resulting in a negative impression [60,61]. Therefore, a more specific and intentional effort is needed to select images that are relevant to the crowdfunding promoter and help viewers understand the current status of the project. This includes ensuring that images are selected that clearly demonstrate the project’s progress, the promoter’s involvement, and key elements related to the crowdfunding goal. When selecting images, consider the quality, relevance, and expressiveness of the images to ensure that they provide viewers with a comprehensive and clear visual representation of the project.

Our study found that the interaction between individual features in crowdfunding images and image themes in crowdfunding visuals will exert different directional influences on the ultimate fundraising success of medical crowdfunding campaigns. For example, themes such as medical and health are positively influenced by an increase in the number of characters in crowdfunding results, while the presence of smiley faces in images for themes such as city life is detrimental to crowdfunding results. This dynamic suggests that images with higher thematic and character complexity may deter potential contributors due to the increased cognitive load required to process them, as suggested by Yang et al [16].

The application of counterfactual analysis effectively helps us to distinguish causal effects from statistical correlations [62-64], providing stronger evidence for understanding which factors directly influence the success of crowdfunding projects. In terms of the performance of the results, the characteristics of the 3 components mentioned above, including certain characteristics of individuals, certification image topic, and the interaction, are all consistent with the results of the regression models. This further confirms the causal relationship between these factors and crowdfunding outcomes.

Selecting an appropriate image theme that aligns with and emphasizes the informational cues of the crowdfunding project may enhance its persuasive appeal [65] and thereby encourage donation behavior. For example, using medical and health–related images with more faces has been shown to increase fundraising effectiveness. This strategy not only increases a campaign’s appeal but also accelerates the achievement of its financial goals.

Analyzing these dynamics can reveal important patterns in crowdfunding results. For example, it can identify which types of projects attract more support or pinpoint image features and themes that are particularly effective in attracting attention and donations. Gathering such insights is critical to refining crowdfunding strategies and developing more effective program designs. These improvements could lead to increased donor engagement and support, thereby improving overall campaign success.

In addition, these findings have practical applications for the design and functionality of crowdfunding platforms. By guiding users to select and upload images that have a higher likelihood of success, such as those depicting younger people or medical scenarios, platforms can strategically influence the visual presentation of campaigns. This targeted approach can optimize the visual impact of these campaigns, potentially increasing their success rates and making the crowdfunding experience more effective for both creators and donors.

For website designers, designers should focus on the selection and presentation of images on the project page to ensure that they are not only clear [66] but also rich in elements that can touch people’s hearts. At the same time, designers can help project sponsors make better choices by creating thematic guidelines or providing a gallery of sample images. Given that different thematic images have different impacts on fundraising results, designers should encourage diversity and provide tools or suggestions to help personalize image content to meet different donor preferences while maintaining overall site consistency and professionalism.

There are certain limitations to our research that open the way for potential avenues of future investigation. First, our analysis focused exclusively on crowdfunding campaigns for individual cancer treatments on GoFundMe within the United States. This intentional scope precluded the examination of cancer treatment crowdfunding projects in other countries, leaving room for future studies to explore the dynamics of such campaigns on a global scale.

Second, our study of image content was limited to character features and image descriptions. To increase the comprehensiveness of future research, it is necessary to delve into more complex aspects of image features. This includes a deeper exploration of the interactive actions depicted in the images to provide a more nuanced understanding of how characters visually engage with each other.

Third, the study’s conclusions are based on currently observed image features but do not sufficiently take into account the dynamic changes in audience preferences. Our study is limited to the analysis of image features, while the deeper psychology of the viewer is not sufficiently explored. There is a lack of in-depth analysis of core factors such as emotion and attention.

Fourth, this study focuses on exploring image associations using a quantitative approach, and future research will combine qualitative methods and take advantage of mixed methods research to explore deeper factors of crowdfunding success.

By addressing these limitations and broadening the scope of image analysis, future research efforts can contribute to a more comprehensive and insightful understanding of crowdfunding dynamics in the context of individual cancer treatments.

In summary, our research examined the influence of image content on the effectiveness of medical crowdfunding initiatives by focusing on specific image features and themes. We rigorously tested 3 hypotheses. Our findings indicate that the representation of individuals in crowdfunding images significantly affects fundraising performance. Certain aspects, such as the presence of multiple individuals, a higher proportion of younger people, and smiles, contribute positively, while a higher proportion of female individuals correlates with a negative impact. The thematic content of the images is also critical, with themes related to hair and beauty, medicine and health, and food and cooking showing positive effects on fundraising. Conversely, themes associated with city life, diverse locations, and rest and play tend to have a negative impact on fundraising efforts. In addition, there is a significant interaction between image themes and the characteristics of the people depicted, influencing the overall fundraising success of the project. This impact varies, increasing or decreasing with the intensity of the image characteristics across themes. Although this study highlights the significant impact of image features on crowdfunding outcomes, viewers' perceptions of images are constantly evolving. Further research into the underlying mechanisms that drive attention in crowdfunding is warranted.