Acceptability of artificial intelligence for cervical cancer screening in Dschang, Cameroon: a qualitative study on patient perspectives

Abstract

Background
Cervical cancer is the fourth most frequent cancer among women, with 90% of cervical cancer-related deaths occurring in low- and middle-income countries like Cameroon. Visual inspection with acetic acid is often used in low-resource settings to screen for cervical cancer; however, its accuracy can be limited. To address this issue, the Swiss Federal Institute of Technology Lausanne and the University Hospitals of Geneva are collaborating to develop an automated smartphone-based image classifier that serves as a computer aided diagnosis tool for cancerous lesions. The primary objective of this study is to explore the acceptability and perspectives of women in Dschang regarding the usage of a screening tool for cervical cancer relying on artificial intelligence. A secondary objective is to understand the preferred form and type of information women would like to receive regarding this artificial intelligence-based screening tool.

Methods
A qualitative methodology was employed to gain better insight into the women’s perspectives. Participants, aged between 30 and 49 were invited from both rural and urban regions and semi-structured interviews using a pre-tested interview guide were conducted. The focus groups were divided on the basis of level of education, as well as HPV status. The interviews were audio-recorded, transcribed, and coded using the ATLAS.ti software.

Results
A total of 32 participants took part in the six focus groups, and 38% of participants had a primary level of education. The perspectives identified were classified using an adapted version of the Technology Acceptance Model. Key factors influencing the acceptability of artificial intelligence include privacy concerns, perceived usefulness, and trust in the competence of providers, accuracy of the tool as well as the potential negative impact of smartphones.

Conclusion
The results suggest that an artificial intelligence-based screening tool for cervical cancer is mostly acceptable to the women in Dschang. By ensuring patient confidentiality and by providing clear explanations, acceptance can be fostered in the community and uptake of cervical cancer screening can be improved.

Trial registration
Ethical Cantonal Board of Geneva, Switzerland (CCER, N°2017–0110 and CER-amendment n°4) and Cameroonian National Ethics Committee for Human Health Research (N°2022/12/1518/CE/CNERSH/SP). NCT: 03757299.

Full text

Sachdevaetal. Reproductive Health (2024) 21:92
https://doi.org/10.1186/s12978-024-01828-8
RESEARCH Open Access
© The Author(s) 2024. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecom‑
mons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Reproductive Health
Acceptability ofarticial intelligence
forcervical cancer screening inDschang,
Cameroon: aqualitative study onpatient
perspectives
Malika Sachdeva1*, Alida Moukam Datchoua2,3, Virginie Flore Yakam2, Bruno Kenfack2,
Magali Jonnalagedda‑Cattin4,5, Jean‑Philippe Thiran4, Patrick Petignat6 and Nicole Christine Schmidt6,7
Abstract
Background Cervical cancer is the fourth most frequent cancer among women, with 90% of cervical cancer‑related
deaths occurring in low‑ and middle‑income countries like Cameroon. Visual inspection with acetic acid is often
used in low‑resource settings to screen for cervical cancer; however, its accuracy can be limited. To address this
issue, the Swiss Federal Institute of Technology Lausanne and the University Hospitals of Geneva are collaborat‑
ing to develop an automated smartphone‑based image classifier that serves as a computer aided diagnosis tool
for cancerous lesions. The primary objective of this study is to explore the acceptability and perspectives of women
in Dschang regarding the usage of a screening tool for cervical cancer relying on artificial intelligence. A secondary
objective is to understand the preferred form and type of information women would like to receive regarding this
artificial intelligence‑based screening tool.
Methods A qualitative methodology was employed to gain better insight into the women’s perspectives. Partici‑
pants, aged between 30 and 49 were invited from both rural and urban regions and semi‑structured interviews using
a pre‑tested interview guide were conducted. The focus groups were divided on the basis of level of education,
as well as HPV status. The interviews were audio‑recorded, transcribed, and coded using the ATLAS.ti software.
Results A total of 32 participants took part in the six focus groups, and 38% of participants had a primary level
of education. The perspectives identified were classified using an adapted version of the Technology Acceptance
Model. Key factors influencing the acceptability of artificial intelligence include privacy concerns, perceived use‑
fulness, and trust in the competence of providers, accuracy of the tool as well as the potential negative impact
of smartphones.
Conclusion The results suggest that an artificial intelligence‑based screening tool for cervical cancer is mostly
acceptable to the women in Dschang. By ensuring patient confidentiality and by providing clear explanations, accept‑
ance can be fostered in the community and uptake of cervical cancer screening can be improved.
*Correspondence:
Malika Sachdeva
mlksachdeva@gmail.com
Full list of author information is available at the end of the article
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 2 of 11
Sachdevaetal. Reproductive Health (2024) 21:92
Trial registration Ethical Cantonal Board of Geneva, Switzerland (CCER, N°2017–0110 and CER‑amendment n°4)
and Cameroonian National Ethics Committee for Human Health Research (N°2022/12/1518/CE/CNERSH/SP). NCT:
03757299.
Plain Language Summary
Globally, cervical cancer is the fourth most frequent cancer among women. However, 90% of all deaths caused
by cervical cancer occur in low‑and middle‑income countries. Methods traditionally used in settings like Cameroon
to detect cervical cancer unfortunately lack accuracy. Therefore, researchers at the Swiss Federal Institute of Technol‑
ogy Lausanne and the University Hospitals of Geneva are developing an artificial intelligence‑based computer aided
diagnosis tool to detect pre‑cancerous lesions using a smartphone application. The aim of this study was to explore
the acceptability and perspectives regarding an AI‑based tool for cervical cancer screening for women in Dschang,
a city in the west of Cameroon. A qualitative methodology was conducted with six focus groups and a total of 32
participants. The main concerns highlighted by the study are related to privacy, trust in the ability of the healthcare
providers, accuracy of the tool as well as the potential negative impact of smartphones. In conclusion, our results
show that a computer aided diagnosis tool using artificial intelligence is mostly acceptable to women in Dschang,
as long as their confidentiality is preserved, and they are provided with clear explanations beforehand.
Keywords Artificial intelligence, Cervical cancer, Patient acceptability, Patient perspectives, Qualitative study
Background
Cervical cancer (CC) represents the fourth most frequent
cancer worldwide among women, with 604,000 new cases
estimated in 2020. However, the global burden of this dis-
ease is unevenly distributed. About 90% of the estimated
342,000 deaths from CC in 2020 occurred in low- and
middle-income countries (LMICs) like Cameroon [1].
Without any significant intervention, the World Health
Organization (WHO) estimates that deaths linked to CC
will increase to 460 000 globally by 2040 with LMICs see-
ing the greatest relative increase [2].
To combat the increasing burden of CC, WHO adopted
a global strategy in November 2020 to accelerate the
elimination of CC as a public health problem and set up
the following “90–70-90” targets that are to be reached
by the year 2030: (i) 90% of girls fully vaccinated by age
15, (ii) 70% of women are screened with a high-perfor-
mance test by 35years, and again by 45 years of age, and,
(iii) 90% of women identified with cervical disease (pre-
cancer and cancer) receive appropriate treatment and
management.
More than 95% of CC cases are linked with a per-
sistent human papillomavirus (HPV) infection. While
CC is highly preventable, with HPV vaccinations, and
screening of precancerous and cancerous lesions, in sub-
Saharan Africa (SSA), lack of CC screening and HPV
vaccination programmes, alongside a high prevalence of
HPV and HIV infections, have contributed to the rising
incidence of CC [3].
In high-income countries (HICs), cytology-based
screening is a mainstay in CC screening [4]. How-
ever, in low-resource settings like Cameroon, WHO
recommends primary HPV testing, followed by visual
inspection with acetic acid (VIA) and Lugol’s iodine
(VILI) for triage of HPV-positive women. VIA is widely
used in LMICs due to its low-cost, even if the sensitiv-
ity varies between 25% to 94.4% due to the subjectivity
of the test depending on the healthcare provider [5].
In 2018, the 3T approach (Test, Triage, Treat),
which takes place as a single visit, has been imple-
mented in Dschang, in collaboration with the Came-
roon Ministry of Public Health, the Dschang Regional
Annex Hospital, and the University Hospitals of
Geneva (HUG) [6]. Primary HPV testing consists of
HPV self-sampling, carried out by the women them-
selves (assisted by midwives if necessary) and analysed
in about an hour (using the GeneXpert®) followed
by VIA-triage for HPV-positive women. Finally, if
needed, treatment by thermal ablation or LEEP coni-
sation can be performed [7].
Even if recent technical developments have improved
early diagnosis of cervical precancerous lesions, accu-
rate visual assessment remains difficult, mainly because
of subjectivity and lack of quality control. Currently,
the Swiss Federal Institute of Technology Lausanne
(EPFL) and HUG are collaborating to develop an auto-
mated smartphone-based image classifier that serves as
a computer aided diagnosis (CAD) tool for cancerous
lesions based on videos obtained only using a smart-
phone application. e images that are recorded are
then classified using an artificial neural network and
image processing techniques that distinguish precan-
cerous and cancerous lesions from non-neoplastic cer-
vical tissue [8]. e results are then available on the
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 3 of 11
Sachdevaetal. Reproductive Health (2024) 21:92
smartphone application to healthcare professionals
and can be shared with patients during consultations to
support explanation and thus improve understanding.
While the use of artificial intelligence (AI) in the health-
care field is increasing in recent years, especially in HICs,
previous studies have evoked various barriers to the
uptake of clinical decision support tools on smartphones
by patients and healthcare providers (HCPs). Concerns
such as theft of devices, fear of a data breach and percep-
tions of reduced patient trust were highlighted, though
more research is needed to better understand the accept-
ability of AI as a clinical decision support tool in patients
in the Global South [9–11].
e aim of this study is to explore the acceptability and
perspectives of females in Dschang, Cameroon, regard-
ing a CAD screening tool for CC relying on AI prior to
its implementation in the clinical setting. A secondary
objective is to understand in which form and content
women would like to receive information about the uti-
lisation of AI for CC screening. Suggestions will be made
accordingly to ensure improved acceptance and under-
standing of AI as a CAD tool for CC screening for future
patients and interventions will be proposed considering
current literature.
Methods
Study site andsetting
e study took place in the Dschang district, Cameroon
in August 2022. e district is composed of Dschang city,
an urban area, and the surrounding rural areas, with a
total population of approximately 220,000 people (Fig.1).
Females that had already participated in the Dschang
Regional Annex Hospital cervical screening programme
(3T) were contacted to share their perspective on the use
of a screening tool for CC relying on AI.
Study design
A qualitative methodology using focus groups (FGs) with
4–7 participants was used to gain better insight into the
perspectives of women in Dschang on the use of AI for
CC screening. While a quantitative methodology allows
for standardised results, it does not capture the nuances
of participant perspectives that can be collected with a
qualitative methodology [13]. Additionally, studies have
demonstrated that qualitative methods promote patient
engagement in research [14].
A pre-tested semi-structured interview guide based
on the current literature on the acceptability of AI and
smartphones in patients and HCPs was used to address
the following categories:
• Pre-existing understanding of CC.
• Usability and acceptability of smartphone in a medi-
cal and personal context.
• Impact of an AI-based diagnosis on patients’ trust in
HCPs.
• Usability and acceptability of asmartphone in a med-
ical and personal context.
Women were given an explanation about the use of an
AI-assisted clinical decision support tool before being
asked to share their perspectives on its use.
Eligibility criteria andparticipant recruitment
Women eligible for the 3T-Approach (30–49 years old)
who had previously participated in the CC screening
program at the Dschang Regional Annex Hospital were
contacted by telephone and invited to participate in this
study. Mainly females from the rural areas of Mbeng and
Fotetsa and the urban area of Fiala-Foreke participated
in the study (Fig.1). e FGs were divided in terms of
the participants’ HPV status and level of education, as
previous studies have reported a correlation between
educational status and health behaviours [15]. Homoge-
neity with respect to education was chosen to ensure a
more free-flowing conversation, where participants did
not feel embarrassed to share their perspectives among
participants of a higher level of education. Additional
socio-demographic characteristics (age, area of residence
and marital status) of all participants were also reported.
Table 1 summarises the key characteristics of the six
focus groups.
Data collection andanalysis
e interviews took place in French and were led by
two Cameroonian anthropologists (ADM and VYF) at
the Dschang Regional Annex Hospital who have been
working on the 3T project since several years. e inter-
views lasted approximately 60 minutes. Compensation
was given to participants to cover their travel costs and
snacks. All FGs were audio-recorded after receiving writ-
ten consent from each participant. ey were subse-
quently transcribed in French, and then analysed using
the ATLAS.ti software (version 22), which allows for data
storage, management, and qualitative analysis. Qualita-
tive content analysis was used to analyse the transcripts,
deductively using a codebook, and complemented by
inductive categories. Selected quotations were then
translated into English.
Technology acceptance model
e Technology Acceptance Model (TAM) is a well-
known and widely used theoretical framework that
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 4 of 11
Sachdevaetal. Reproductive Health (2024) 21:92
aims to explain the various factors influencing an indi-
vidual’s acceptance and use of technology, notably their
perceived usefulness of the technology and perceived
ease-of-use. Positive acceptance of these factors is cor-
related with a positive intention resulting in the actual
use of the technology. e model has been studied in
various contexts, including healthcare settings, as well
as in different target populations, leading to various
adaptations and augmentations. In their 2020 study,
Dhagarra etal. propose an extended version of TAM
by additionally integrating trust and privacy concerns
as predictors of patients’ acceptance of technology in
healthcare delivery settings [16]. at revised frame-
work will be used for further analysis of the patients’
perspectives in the current study. Figure2 depicts the
original model, alongside the modifications proposed
by Dhagarra etal. in grey.
Fig. 1 Map of the health areas of the District of Dschang, West Cameroon, modified from Ministère de la Santé Publique du Cameroun (https://
dhis‑ minsa nte‑ cm. org/ portal/), used from a publication with permission of Datchoua Moukam A.M. [12].
Table 1 Description of the main characteristics of the
participants in all six focus groups
Group number HPV Status Level of Education
1Positive; untreated Primary
2Secondary/Superior
3Positive; treated Primary
4Secondary/Superior
5Negative Primary
6Secondary/Superior
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 5 of 11
Sachdevaetal. Reproductive Health (2024) 21:92
Results
Socio‑demographic characteristics
e six FGs consisted of a total of 32 femaleparticipants,
aged between 30 and 48 years old. A primary level of educa-
tion was observed in 38% of the participants, while 40% had
a secondary level, and 22% had a tertiary level of education.
Most women (94%) claimed to either be in a relationship or
married. Finally, 72% of all participants had used a smart-
phone at least once in their lifetime. Table2 summarizes the
socio-demographic characteristics of the participants.
Perceived usefulness andease‑of‑use
As described in Fig.2, perceived usefulness and ease-of-
use are two key components of the original TAM. Per-
ceived usefulness can be defined as “the degree to which
a person believes that using a particular system would
enhance his or her job performance” [17]. From a patient’s
perspective, perceived usefulness can also be described as
the benefits they expect of the technology. Perceived ease-
of-use is defined as “the degree to which a person believes
that using a particular system would be free of effort” [16].
Perceived ease of use was not mentioned specifically
by the participants. However, participants in all FGs
underlined the usefulness of the application, especially
in terms of increased efficiency, precision in diagnosis,
and facilitation of communication.
“We say that it [the computer aided diagnosis tool]
is good, it facilitates the nurses’ work since the eye
cannot visualise well; the device is there to visualise
better and it is very fast….”
(Participant 3, FG2)
Participants also recognised that the use of smart-
phones facilitated communication between providers, as
well as patient-provider communication, since it allowed
them to visualise their own cervix and any potential
lesions after the gynaecological exam.
Furthermore, two additional factors emerged: trust and
privacy concerns. ose were also proposed by Dhagarra
etal. [16] in their study and our findings from Cameroon
are described in the following paragraph.
Trust
ree aspects of trust were discussed during the FGs: trust
in the accuracy of an AI-assisted diagnosis, trust in the safety
of the procedure and trust in the competency of HCPs.
When asked about their level of trust in the accu-
racy of a diagnostic made using AI, the responses of the
Fig. 2 An adapted version of the TAM with the augmentations proposed by Dhagarra et al. in grey
Table 2 Socio‑demographic characteristics of participants
Characteristics Frequency (n) %
Level of Education
Primary education 12 38
Secondary education 13 40
Tertiary education 7 22
Marital status
Married 20 63
Divorced 1 3
In a relationship 10 31
Single 1 3
Use of smartphones at least once
Yes 23 72
No 9 28
Age (years)
Range: 30–48
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 6 of 11
Sachdevaetal. Reproductive Health (2024) 21:92
participants varied from 50 to 100%. Lack of complete
trust in the diagnosis was often attributed to the fact
that the system is not necessarily 100% error-free and
can malfunction. e competence of HCPs while film-
ing using the smartphone was also evoked as a factor that
could influence the accuracy of the diagnosis.
“I would say 80% in favor because a device can have
problems, so we must be sure that the device is in
good condition and there is also the user; does the
user use it properly? Yes, because for example, if I
am a nonprofessional and you give me this, I will
take and film as I want! So, you need qualified per-
sonnel for that, yes, you need someone qualified.”
(Participant 2, FG2).
Lack of trust in the safety of smartphones was another
important aspect that was highlighted by three out of 32
participants in two different FGs. Participant concerns
were related to the dangerous effect smartphones could
have on their own health.
“If I had tested positive, I would not want to be cap-
tured by a telephone because I would tell myself that
either it aggravates my cancer, or…actually I would
not like to be captured by a telephone. Maybe if it
were with an ultrasound, but with the phone, know-
ing that they emit radiations, I would not want to be
filmed by a telephone.”
(Participant 1, FG 5).
Finally, two participants questioned the impact of
dependence on the smartphone application on the com-
petency of HCPs. One participant expressed concerns
about HCPs being overly reliant on the smartphone
and another expressed that HCPs should be away from
smartphones while working since they are an important
source of distraction.
Privacy concerns
Privacy and data protection concerns were a significant point
of discussion in each of the six FGs and different perspectives
were mentioned. Less than a third of participants, especially
smartphone users, feared that their images could potentially
be published on social media pages, like Facebook.
“A risk is the protection of data of the patients
mainly. Yes! You keep them in the phone, you..you
guarantee protection, but it is a phone. Artificial
intelligence or not, once it is connected, data is no
longer protected from what I know.”
(Participant 2, Focus Group 4)
Most participants were less concerned about the pro-
tection of their data and believed that even if their photos
were published, their identities would remain protected.
e absence of these concerns was closely related to con-
fidence in HCPs, highlighting that upholding patient con-
fidentiality was the HCP’s responsibility.
“I too have trust since in this profession, confidenti-
ality is a must and I think that you truly have to be
someone with a certain immorality to publish such
things. I have confidence. Since the beginning when
they spoke to me of this, I accepted.”
(Participant 2, FG2)
Informational needs
e secondary objective of this study was to understand the
type and form of information women would like to receive
about the utilisation of AI for CC screening. About a third
of women were satisfied to know their result of the HPV-
test and did not require further information about the use of
AI, as explained by a woman from a secondary focus group:
“For me, nothing [no information] in particular. I
am just waiting for the results, that they first have a
look at the photos with the old method [without AI],
that they perhaps know if there really are or not the
lesions, and that we then use this artificial [intelli-
gence] method, and if all is still good, I am ok. I am
just waiting for the results, nothing else.”
(Participant 1, FG4).
On the other hand, a few women wanted to better
understand how the smartphone application worked and
whether they themselves could use it.
“I would just like to know the specifics, actually
understand the artificial intelligence, what it is
going to do....what is artificial intelligence. For exam-
ple, when we talk about an application, we give its
advantages, its limits, how it leads to the result that
we expect […] Just give some understandable infor-
mation about the app itself. So that I know what
the artificial intelligence that analyzed my data is
called, how it reacts. It is always a matter of trust
and knowledge too.”
(Participant 2, FG4).
With respect to the form of information, participants
in all six FGs stated that they would prefer a face-to-face
conversation with HCPs.
“Yes, a conversation face-to-face because when you
discuss with her [the HCP], you can ask questions
and she can respond.”
(Participant 2, FG1).
In addition, a few women mentioned that they would
benefit from an additional explicative video or a written
explanation, as a brochure.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 7 of 11
Sachdevaetal. Reproductive Health (2024) 21:92
“I prefer a brochure and an explanation face-to-face.
A brochure so that I can keep the information for
myself and when I am not at the hospital, I can still
inform myself and research what is written. And the
explanation because there are little subtilities that
the person that explains knows best and it perhaps
also allows us to ask questions on what is written so
for me the two methods should be used.”
(Participant 2, FG4).
Inuence ofeducational status onthefactors inuencing
theacceptance ofAI‑based diagnostics
As the FGs were organized according to the partici-
pant’s educational level, the influence of education on
the previously described factors was explored. Most
women of primary and superior levels of education
saw the tool as an advanced use of technology and were
curious about its use.
“We are very grateful for the fact that science is
making progress. When it does, it means that we
can quickly find our cure!”
(Participant 1, FG1).
Furthermore, they trusted the CAD tool, mainly because
of its increasing precision. However, even if a minority of
women mainly with a primary level of education stated it
was a “Western influence” that they were unable to under-
stand, women in the FGs with a primary education level
mainly expressed their confidence in the tool.
“We have complete trust if they don’t publish [our
photos] in any which way.”
(Participant 2, FG1).
Overall, while the barriers mentioned by both groups
were similar, including confidentiality concerns and
trust, women with a higher educational status exhib-
ited a greater tendency to question the reliability of the
tool. Hereby, women with a higher level of education,
who were also more experienced smartphone users,
discussed concerns regarding potential smartphone
malfunctions, the impreciseness of the tool and the
inexperience of HCPs.
Discussion
e following discussion section will primarily address
key findings that are important for HCPs to consider,
when using a smartphone including AI-aided CAD tools,
to avoid negative impacts on the uptake of CC screening
by the women or on return to follow-up. In this study,
the acceptance of women in Dschang regarding a CAD
tool for CC relying on AI was studied. e TAM was
identified as an appropriate model to investigate the vari-
ous factors influencing the acceptance of the AI-based
diagnostic aid. We identified that, in the selected study
setting, patients’ acceptance was mainly impacted by per-
ceived usefulness, privacy concerns and trust. Perceived
ease-of-use was not a main concern probably because
patients are not end-users of the application themselves
and so are unaffected by its usability. e various themes
that were revealed during the study were categorized
according to the TAM, as depicted in blue in Fig.3.
Overall participants perceived the AI-based CAD tool
on a smartphone as a support for enhancing the diagno-
ses of CC. e findings can be compared in theory on
two distinct levels: first, the use of the smartphones to
diagnose CC and second, the use of AI to diagnose the
disease. However, many participants in the study made
no difference between the use of smartphones and an
AI-based technology. Nevertheless, the following section
will highlight important factors and differences between
Fig. 3 Summary of the factors explored during the study (denoted in blue) using the modified TAM
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 8 of 11
Sachdevaetal. Reproductive Health (2024) 21:92
their perceptions related to the use of smartphones and
AI-based technology when possible.
Acceptability ofsmartphones
Regarding the acceptability of smartphones, a study by
Mungo etal., showed that smartphone-based cervicog-
raphy is highly acceptable in HPV-positive women liv-
ing with HIV in Western Kenya [17]. is observation
was confirmed in our study, but it is important to point
out that a few women expressed apprehension about the
radiation emitted by smartphones and thus were less
inclined to accept the method. Similar findings were
encountered in another study in Kenya that analysed
HCPs and patients’ perspectives on using an mHealth
tool for eye care [10]. Indeed, in this study, a minority
of patients were concerned about the negative health
impact of mobile phones and therefore preferred tradi-
tional methods instead. Even if this perception was not
shared by most women in our settings, it is important
that HCPs are aware of this barrier. In a study aiming to
understand and prevent health concerns about emerging
mobile health technologies, Materia etal. highlighted the
importance of evaluating and addressing health concerns
related to use of smartphones and developing evidence-
based communication strategies to limit them [18].
Professionals, therefore, need to be trained adequately
to address safety concerns, including misconceptions,
prior to the implementation of new technologies. Effec-
tive patient-centred communication can help recognise
and address these concerns. Moreover, misconceptions
especially need to be explored in future studies to better
understand their potential impact on CC screening.
Acceptability ofAI‑based CC screening
Since AI-based CC screening is a novel method, there are
limited studies concerning the acceptability of this tech-
nique by patients, especially in LMICs. In this study, a
few women, especially those with a higher level of edu-
cation, exhibited reservations about the AI-based tool’s
potential to malfunction (either due to issues in the
smartphone or the impreciseness of the AI). Some also
expressed concerns of the inexperience of the HCP han-
dling the tool. However, most women viewed the AI-
based tool as an enhanced method to improve diagnostic
precision of HCPs.
Firstly, it needs to be acknowledged that our results are
in line with the current literature on the use of AI in the
field of cancer diagnosis. Studies report that patients tend
to accept AI more easily if a dangerous disease such as
cancer can be avoided [19]. Other studies exploring the
perceptions of AI in the diagnosis of breast and skin can-
cer concluded that it was an acceptable adjunctive tech-
nology for patients, if the provider-patient relationship
was preserved, but not an acceptable substitute for physi-
cians or radiologists [15, 20, 21]. It is important to note,
however, that these studies were conducted in high-
income countries, i.e. the United Kingdom, United States,
and Italy and no studies in SSA could be identified.
With respect to HCP competence concerns encoun-
tered in our study, similar perspectives were reported
by other studies exploring the acceptability of similar
CAD tools in various healthcare domains. A study in
Uganda regarding HCPs’ perspectives on the acceptabil-
ity of a mobile health tool revealed a concern that using
a mobile application in front of patients and their fami-
lies would undermine their trust in the HCP’s ability to
diagnose and treat [9]. is perception was echoed in
another study in the UK where HCPs were worried that
using mobile tools during patient interactions would be
perceived as unprofessional [11]. However, our findings
demonstrate that if the usage of smartphones is explained
to the patients beforehand, most participants had con-
tinued trust and confidence in the HCPs as well as the
diagnosis made by the AI-based tool. Only a minority of
participants thought that the usage of this smartphone
application would make HCPs overly reliant, distracted,
or lazier. Nonetheless, HCPs need to be adequately
trained on how to communicate the inclusion of the AI-
device during patient consultations, to avoid mispercep-
tions regarding its use.
Importantly, regarding communication, most partici-
pants recognized that smartphones are an important tool
to facilitate communication between HCPs and patients.
By showing the images of their cervix to patients, HCPs
can promote transparency and thus establish trust and
reinforce patient-provider communication.
Patient condentiality
e last factor that is important to address is privacy
and confidentiality concerns, which were revealed across
all groups regardless of their education level. Concerns
regarding videos being posted on social media were
highlighted, especially when smartphones are used by
students or trainees. ese concerns have been under-
lined in previous studies, such as a meta-synthesis of
qualitative studies evaluating public perceptions of AI in
healthcare, mostly conducted in the Global North [22].
However, these concerns were highlighted to a lesser
extent in studies based in the Global South. A study
based in Bangladesh reported that privacy concerns had
an insignificant impact on the adoption of eHealth [23].
Similarly, a study by Asgary etal. on the CC screening
with smartphones in Ghana reported minimal privacy
concerns [24]. Finally, in a systematic review analysing
barriers to the use of mobile health in developing coun-
tries, privacy and confidentiality concerns were evoked
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 9 of 11
Sachdevaetal. Reproductive Health (2024) 21:92
only 3% of the time [25]. However, HCPs should address
these concerns by explaining to the patients procedures
of data storage and handling of data protection issues.
In summary, to address factors affecting the acceptance
of either smartphones or AI-based technology, patients
need to be informed adequately by HCPs. Hereby, par-
ticipants unanimously agreed that an explanation face-
to-face with the HCPs would be the best way to provide
information regarding the AI-based tool, as it would
allow them to ask questions and receive the information
that they need. Additionally, it should be acknowledged
that some participants (especially those with higher edu-
cation) were interested in additional means of informa-
tion, such as brochures or videos.
Strengths andlimitations
Although this study is, to the best of our knowledge, one of
the first studies in Cameroon that analyses the acceptabil-
ity of AI for CC screening and one of the few in SSA, some
limitations need to be acknowledged. First, an interviewer or
selection bias cannot be excluded. As some FGs were con-
ducted in larger groups of 7 participants, participants’ ability
to express their opinions freely might be affected. To mitigate
this, FGs were limited to women of similar educational back-
grounds and were conducted by a Cameroonian anthropolo-
gist. However, the anthropologist’s higher level of education
may have influenced participants’ responses, especially in
FGs of participants with a primary level of education.
Moreover, even if the FGs were comprised of women
of diverse socio-economic backgrounds, the participants
had already taken part in the CC screening program and
therefore may have systematically included those who are
inclined towards prevention and have already been sen-
sitized to the importance of CC screening. Furthermore,
the study revealed that the acceptability of an AI-based
tool can be influenced by the utilisation of smartphones,
highlighting the need to explore differences affecting
acceptability in larger qualitative and quantitative studies.
Additionally, the use of the TAM can be seen as simplis-
tic, and its suitability can be debated, since it is primar-
ily intended for end-users of the technology, which the
patients were not in our study. However, in our opinion,
using the TAM allowed us to classify and understand
existing barriers and facilitators to AI-based diagnostics.
e last limitation can be seen in the methodology of
the study. A qualitative methodology was chosen as an
appropriate approach since it allowed us to explore the
perspectives of females and capture insights into the ways
people perceive and interpret their surroundings [14, 26].
Butqualitative studies have limited generalizability. How-
ever, as saturation was achieved for most themes such as
privacy concerns and perceived usefulness, we consider
the results of the study to be important for similar settings.
Given the strengths and limitations of our study, further
research is needed. Hereby, the following three areas of
research seem important. Firstly, to gain a broader insight
into patients’ acceptability of AI, a quantitative analysis
should be considered exploring the influence of women’s
educational status, but also including the perspectives of
spouses and other community members, since they have a
significant impact on patients’ attitudes. Secondly, miscon-
ceptions about the dangers of smartphones in the commu-
nity should be explored. Finally, acceptability should also
be assessed after an initial pilot phase of the AI tool in CC
screening, since perspectives may shift after the experience.
Conclusions
Overall, our findings suggest that an AI-assisted screen-
ing tool for CC can therefore be seen as largely accept-
able to women in Dschang, irrespective of their
socio-economic status and level of education. However,
acceptability is significantly contingent on preserving the
confidentiality and privacy of the images or videos taken.
We therefore recommend that all participants receive
a counselling session before the AI-based screening that
informs the patients about the steps of the procedure, the
purpose, and advantages of the tool, as well as potential
risks that are associated with it. Explanations related to how
the photos and videos will be stored should be provided as
well as an assurance that the images taken will solely be for
the use of other HCPs. It would also provide the patients
with an opportunity to ask questions if needed.
HCPs, therefore, need to be trained in effective patient-
centred communication. By ensuring patient confidenti-
ality and by providing clear explanations, acceptance of
this method can be fostered in the community, thereby
improving the uptake of CC screening, and reducing the
burden of CC in LMICs.
In summary, based on the results of our study, the fol-
lowing three suggestions can be made to HCPs when they
introduce this CAD tool for CC screening to patients:
1. Emphasize that the AI-assisted tool is used to pro-
vide information about and to assist the diagnosis.
2. Ensure the protection of patients’ data and provide the
patients with assurance regarding its strict confidentiality.
3. Convey the purpose, benefits, and potential risks of
the tool in written form (brochures) for patients who
would like to learn more about the tool.
is AI-assisted diagnosis tool could improve CC
screening in Dschang and therefore reduce the burden of
this women’s health concern in the region. However, the
successful implementation of this tool hinges on accept-
ance from both HCPs and patients. is study indicates
that patients tend to be open to AI-based solutions.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 10 of 11
Sachdevaetal. Reproductive Health (2024) 21:92
Abbreviations
CC Cervical cancer
LMIC Low‑and‑middle income country
WHO World Health Organization
HPV Human papillomavirus
SSA Sub‑Saharan Africa
HIC High‑income country
VIA/VILI Visual inspection with acetic acid/ Lugol’s iodine
HUG University hospitals of Geneva
EPFL Swiss federal institute of technology lausanne
CAD Computer‑aided diagnosis
AI Artificial intelligence
HCP Healthcare provider
FG Focus group
TAM Technology acceptance model
Acknowledgements
We are immensely grateful to all study participants for their time and invalu‑
able contributions to this research.
Authors’ contributions
MS supported all phases of the research and wrote the first draft of the
manuscript. ADM & VFY were responsible for the recruitment of all study par‑
ticipants and data collection. BK supervised the study on‑site. MJC & JPT took
part in the conceptualisation of the study and revised the paper. PP assisted
and supervised the conception of the study and the writing. NCS helped in all
phases of the research and revised the paper. All authors read and approved
the final version.
Funding
Open access funding provided by University of Geneva This work was sup‑
ported by funding received from Tech4Dev. The funders had no role in the
study design, data collection and analysis, decision to publish, or preparation
of the manuscript.
Availability of data and materials
Datasets (transcripts) are not publicly available due to the sensitivity of the
data, but the interview guide or summaries of transcripts (including catego‑
ries and codes) can be made available from the corresponding author upon
reasonable request.
Declarations
Ethics approval and consent to participate
The study is part of a larger trial, which was approved by the Ethical Cantonal
Board of Geneva, Switzerland (CCER, N°2017–0110 and CER‑amendment n°4)
and the Cameroonian National Ethics Committee for Human Health Research
(N°2022/12/1518/CE/CNERSH/SP).
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Author details
1 Faculty of Medicine, University of Geneva, Geneva, Switzerland. 2 Depart‑
ment of Gynaecology and Obstetrics, Dschang Regional Annex Hospital,
Dschang, Cameroon. 3 Institute of Global Health, Faculty of Medicine, Uni‑
versity of Geneva, Geneva, Switzerland. 4 Signal Processing Laboratory LTS5,
School of Engineering, École Polytechnique Fédérale de Lausanne, Lausanne,
Switzerland. 5 EssentialTech Centre, École Polytechnique Fédérale de Lausanne,
Lausanne, Switzerland. 6 Gynaecology Division, Department of Paediat‑
rics, Gynaecology and Obstetrics, University Hospitals of Geneva, Geneva,
Switzerland. 7 Faculty of Social Science, Catholic University of Applied Science,
Munich, Germany.
Received: 27 August 2023 Accepted: 12 June 2024
References
1. Cervical cancer. https:// www. who. int/ news‑ room/ fact‑ sheets/ detail/ cervi
cal‑ cancer. Accessed 29 July 2022.
2. Cervical cancer. In: WHO | Regional Office for Africa. https:// www. afro.
who. int/ health‑ topics/ cervi cal‑ cancer. Accessed 1 Aug 2022.
3. Jedy‑Agba E, Joko WY, Liu B, et al. Trends in cervical cancer incidence in
sub‑Saharan Africa. Br J Cancer. 2020;123:148–54.
4. Indicator Metadata Registry Details. https:// www. who. int/ data/ gho/ indic
ator‑ metad ata‑ regis try/ imr‑ detai ls/ 3240 .Accessed 9 Aug 2023.
5. Viñals R, Jonnalagedda M, Petignat P, Thiran JP, Vassilakos P. Artificial
Intelligence‑Based Cervical Cancer Screening on Images Taken during
Visual Inspection with Acetic Acid: ASystematic Review. Diagnostics
(Basel). 2023;13(5):836. https:// doi. org/ 10. 3390/ diagn ostic s1305 0836.
6. Vassilakos P, Tebeu P‑M, Halle‑Ekane GE, Sando Z, Kenfack B, Baumann F,
Petignat P. Vingt années de lutte contre le cancer du col utérin en Afrique
subsaharienne ‑ Collaboration médicale entre Genève et Yaoundé. Rev
Med Suisse. 2019;642:601–5.
7. Grohar D, Vassilakos P, Benkortbi K, Tincho E, Kenfack B, Petignat P. Scaling
up community‑based cervical cancer screening in Cameroon employing
a single visit approach. International Journal of Gynecologic Cancer. 2020.
https:// doi. org/ 10. 1136/ ijgc‑ 2020‑ 001422.
8. Baleydier I, Vassilakos P, Viñals R, et al. Study protocol for a two‑site clinical
trial to validate a smartphone‑based artificial intelligence classifier identi‑
fying cervical precancer and cancer in HPV‑positive women in Cameroon.
PLoS ONE. 2021;16: e0260776.
9. Ellington LE, Najjingo I, Rosenfeld M, et al. Health workers’ perspectives of
a mobile health tool to improve diagnosis and management of paediat‑
ric acute respiratory illnesses in Uganda: a qualitative study. BMJ Open.
2021;11: e049708.
10. Lodhia V, Karanja S, Lees S, Bastawrous A. Acceptability, Usability, and Views
on Deployment of Peek, a Mobile Phone mHealth Intervention for Eye Care
in Kenya: Qualitative Study. JMIR Mhealth Uhealth. 2016;4: e4746.
11. Patel R, Green W, Shahzad MW, Larkin C. Use of Mobile Clinical Decision Sup‑
port Software by Junior Doctors at a UK Teaching Hospital: Identification and
Evaluation of Barriers to Engagement. JMIR Mhealth Uhealth. 2015;3: e4388.
12. Datchoua Moukam AM, Embolo Owono MS, Kenfack B, Vassilakos P,
Petignat P, Sormani J, Schmidt NC. "Cervical cancer screening: awareness
is not enough". Understanding barriers to screening among women in
West Cameroon‑a qualitative study using focus groups. Reprod Health.
2021;18(1):147. https:// doi. org/ 10. 1186/ s12978‑ 021‑ 01186‑9.
13. Gibbons C, Singh S, Gibbons B, Clark C, Torres J, Cheng MY, Wang EA,
Armstrong AW. Using qualitative methods to understand factors contrib‑
uting to patient satisfaction among dermatology patients: a systematic
review. J Dermatol Treat. 2018;29:290–4.
14. Rolfe DE, Ramsden VR, Banner D, Graham ID. Using qualitative Health
Research methods to improve patient and public involvement and
engagement in research. Res Involv Engagem. 2018;4:49.
15. Pesapane F, Rotili A, Valconi E, et al. Women’s perceptions and attitudes
to the use of AI in breast cancer screening: a survey in a cancer referral
centre. BJR. 2023;96:20220569.
16. Dhagarra D, Goswami M, Kumar G. Impact of Trust and Privacy Concerns
on Technology Acceptance in Healthcare: An Indian Perspective. Int J
Med Informatics. 2020;141: 104164.
17. Mungo C, Osongo CO, Ambaka J, Randa MA, Samba B, Ochieng CA,
Barker E, Guliam A, Omoto J, Cohen CR. Feasibility and Acceptability
of Smartphone‑Based Cervical Cancer Screening Among HIV‑Positive
Women in Western Kenya. JCO Glob Oncol. 2021;7:686–93.
18. Materia FT, Faasse K, Smyth JM. Understanding and Preventing Health
Concerns About Emerging Mobile Health Technologies. JMIR Mhealth
Uhealth. 2020;8: e14375.
19. McCradden MD, Sarker T, Paprica PA. Conditionally positive: a qualitative
study of public perceptions about using health data for artificial intel‑
ligence research. BMJ Open. 2020;10: e039798.
20. Lim K, Neal‑Smith G, Mitchell C, Xerri J, Chuanromanee P. Perceptions of
the use of artificial intelligence in the diagnosis of skin cancer: an outpa‑
tient survey. Clin Experimental Derm. 2022;47:542–6.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 11 of 11
Sachdevaetal. Reproductive Health (2024) 21:92
21. Nelson CA, Pérez‑Chada LM, Creadore A, et al. Patient Perspectives on the
Use of Artificial Intelligence for Skin Cancer Screening. JAMA Dermatol.
2020;156:1–12.
22. Wu C, Xu H, Bai D, Chen X, Gao J, Jiang X. Public perceptions on the appli‑
cation of artificial intelligence in healthcare: a qualitative meta‑synthesis.
BMJ Open. 2023;13: e066322.
23. Hoque MR, Bao Y, Sor war G. Investigating factors influencing the adop‑
tion of e‑Health in developing countries: A patient’s perspective. Inform
Health Soc Care. 2017;42:1–17.
24. Asgary R, Cole H, Adongo P, Nwameme A, Maya E, Adu‑Amankwah
A, Barnett H, Adanu R. Acceptability and implementation challenges
of smartphone‑based training of community health nurses for visual
inspection with acetic acid in Ghana: mHealth and cervical cancer
screening. BMJ Open. 2019;9: e030528.
25. Kruse C, Betancourt J, Ortiz S, Valdes Luna SM, Bamrah IK, Segovia N.
Barriers to the Use of Mobile Health in Improving Health Outcomes in
Developing Countries: Systematic Review. J Med Internet Res. 2019;21:
e13263.
26. Morgan DL. The SAGE Handbook of Interview Research: The Complexity
of the Craft. 2012. https:// doi. org/ 10. 4135/ 97814 52218 403.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in pub‑
lished maps and institutional affiliations.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com