Using artificial intelligence on dermatology conditions in Uganda: a case for diversity in training data sets for machine learning
Background: In pursuit of applying universal non-biased Artificial Intelligence (AI) in healthcare, it is essential that data from different geographies are represented. Objective: To assess the diagnostic performance of an AI-powered dermatological algorithm called Skin Image Search on Fitzpatrick6...
Gespeichert in:
| Veröffentlicht in: | African health sciences Jg. 23; H. 2; S. 753 - 63 |
|---|---|
| Hauptverfasser: | , , , , , , , |
| Format: | Journal Article |
| Sprache: | Englisch |
| Veröffentlicht: |
Uganda
Makerere Medical School
01.06.2023
|
| Schlagworte: | |
| ISSN: | 1680-6905, 1729-0503, 1729-0503 |
| Online-Zugang: | Volltext |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Abstract | Background: In pursuit of applying universal non-biased Artificial Intelligence (AI) in healthcare, it is essential that data from different geographies are represented.
Objective: To assess the diagnostic performance of an AI-powered dermatological algorithm called Skin Image Search on Fitzpatrick6 skin type (dark skin) dermatological conditions.
Methods: 123 dermatological images selected from a total of 173 images were retrospectively extracted from the electronic database of a Ugandan telehealth company, The Medical Concierge Group (TMCG) after getting their consent. Details of age, gender, and dermatological clinical diagnosis were analysed using R on R studio software to assess the diagnostic accuracy of the AI app along with disease diagnosis and body part. Predictability levels of the AI app were graded on a scale of 0 to 5, where 0- no prediction was made and 1-5 demonstrated a reduction incorrect diagnosis prediction rate of the AI.
Results: 76 (62%) of the dermatological images were from females and 47 (38%) from males. Overall diagnostic accuracy of the AI app on black dermatological conditions was low at 17% (21 out of 123 predictable images) compared to 69.9% performance on Caucasian skin type as reported from the training results. There were varying predictability levels correctness i.e., 1-8.9%, 2-2.4%, 3-2.4%, 4-1.6%, 5-1.6% with performance along individual diagnosis highest with dermatitis (80%).
Conclusion: There is need for diversity of image datasets used to train dermatology algorithms for AI applications to increase accuracy across skin types and geographies.
Keywords: Artificial intelligence; Dermatology; Fitzpatrick 6 skin type; Telehealth; Algorithms. |
|---|---|
| AbstractList | In pursuit of applying universal non-biased Artificial Intelligence (AI) in healthcare, it is essential that data from different geographies are represented.BackgroundIn pursuit of applying universal non-biased Artificial Intelligence (AI) in healthcare, it is essential that data from different geographies are represented.To assess the diagnostic performance of an AI-powered dermatological algorithm called Skin Image Search on Fitzpatrick 6 skin type (dark skin) dermatological conditions.ObjectiveTo assess the diagnostic performance of an AI-powered dermatological algorithm called Skin Image Search on Fitzpatrick 6 skin type (dark skin) dermatological conditions.123 dermatological images selected from a total of 173 images were retrospectively extracted from the electronic database of a Ugandan telehealth company, The Medical Concierge Group (TMCG) after getting their consent. Details of age, gender, and dermatological clinical diagnosis were analysed using R on R studio software to assess the diagnostic accuracy of the AI app along with disease diagnosis and body part. Predictability levels of the AI app were graded on a scale of 0 to 5, where 0- no prediction was made and 1-5 demonstrated a reduction incorrect diagnosis prediction rate of the AI.Methods123 dermatological images selected from a total of 173 images were retrospectively extracted from the electronic database of a Ugandan telehealth company, The Medical Concierge Group (TMCG) after getting their consent. Details of age, gender, and dermatological clinical diagnosis were analysed using R on R studio software to assess the diagnostic accuracy of the AI app along with disease diagnosis and body part. Predictability levels of the AI app were graded on a scale of 0 to 5, where 0- no prediction was made and 1-5 demonstrated a reduction incorrect diagnosis prediction rate of the AI.76 (62%) of the dermatological images were from females and 47 (38%) from males. Overall diagnostic accuracy of the AI app on black dermatological conditions was low at 17% (21 out of 123 predictable images) compared to 69.9% performance on Caucasian skin type as reported from the training results. There were varying predictability levels correctness i.e., 1-8.9%, 2-2.4%, 3-2.4%, 4-1.6%, 5-1.6% with performance along individual diagnosis highest with dermatitis (80%).Results76 (62%) of the dermatological images were from females and 47 (38%) from males. Overall diagnostic accuracy of the AI app on black dermatological conditions was low at 17% (21 out of 123 predictable images) compared to 69.9% performance on Caucasian skin type as reported from the training results. There were varying predictability levels correctness i.e., 1-8.9%, 2-2.4%, 3-2.4%, 4-1.6%, 5-1.6% with performance along individual diagnosis highest with dermatitis (80%).There is need for diversity of image datasets used to train dermatology algorithms for AI applications to increase accuracy across skin types and geographies.ConclusionThere is need for diversity of image datasets used to train dermatology algorithms for AI applications to increase accuracy across skin types and geographies. In pursuit of applying universal non-biased Artificial Intelligence (AI) in healthcare, it is essential that data from different geographies are represented. To assess the diagnostic performance of an AI-powered dermatological algorithm called Skin Image Search on Fitzpatrick 6 skin type (dark skin) dermatological conditions. 123 dermatological images selected from a total of 173 images were retrospectively extracted from the electronic database of a Ugandan telehealth company, The Medical Concierge Group (TMCG) after getting their consent. Details of age, gender, and dermatological clinical diagnosis were analysed using R on R studio software to assess the diagnostic accuracy of the AI app along with disease diagnosis and body part. Predictability levels of the AI app were graded on a scale of 0 to 5, where 0- no prediction was made and 1-5 demonstrated a reduction incorrect diagnosis prediction rate of the AI. 76 (62%) of the dermatological images were from females and 47 (38%) from males. Overall diagnostic accuracy of the AI app on black dermatological conditions was low at 17% (21 out of 123 predictable images) compared to 69.9% performance on Caucasian skin type as reported from the training results. There were varying predictability levels correctness i.e., 1-8.9%, 2-2.4%, 3-2.4%, 4-1.6%, 5-1.6% with performance along individual diagnosis highest with dermatitis (80%). There is need for diversity of image datasets used to train dermatology algorithms for AI applications to increase accuracy across skin types and geographies. Background: In pursuit of applying universal non-biased Artificial Intelligence (AI) in healthcare, it is essential that data from different geographies are represented. Objective: To assess the diagnostic performance of an AI-powered dermatological algorithm called Skin Image Search on Fitzpatrick6 skin type (dark skin) dermatological conditions. Methods: 123 dermatological images selected from a total of 173 images were retrospectively extracted from the electronic database of a Ugandan telehealth company, The Medical Concierge Group (TMCG) after getting their consent. Details of age, gender, and dermatological clinical diagnosis were analysed using R on R studio software to assess the diagnostic accuracy of the AI app along with disease diagnosis and body part. Predictability levels of the AI app were graded on a scale of 0 to 5, where 0- no prediction was made and 1-5 demonstrated a reduction incorrect diagnosis prediction rate of the AI. Results: 76 (62%) of the dermatological images were from females and 47 (38%) from males. Overall diagnostic accuracy of the AI app on black dermatological conditions was low at 17% (21 out of 123 predictable images) compared to 69.9% performance on Caucasian skin type as reported from the training results. There were varying predictability levels correctness i.e., 1-8.9%, 2-2.4%, 3-2.4%, 4-1.6%, 5-1.6% with performance along individual diagnosis highest with dermatitis (80%). Conclusion: There is need for diversity of image datasets used to train dermatology algorithms for AI applications to increase accuracy across skin types and geographies. Keywords: Artificial intelligence; Dermatology; Fitzpatrick 6 skin type; Telehealth; Algorithms. |
| Author | Rusoke, Davis Musinguzi, Davis Bwanika, John Nassiwa, Faith Lubega, William Börve, Alexander Okello, Mark Kamulegeya, Louis |
| AuthorAffiliation | 1 The Medical Concierge Group, Research and Projects 2 The Medical Concierge Group, Information and Technology 3 Sahlgrenska University Hospital, Departments of Orthopaedics |
| AuthorAffiliation_xml | – name: 1 The Medical Concierge Group, Research and Projects – name: 3 Sahlgrenska University Hospital, Departments of Orthopaedics – name: 2 The Medical Concierge Group, Information and Technology |
| Author_xml | – sequence: 1 givenname: Louis surname: Kamulegeya fullname: Kamulegeya, Louis – sequence: 2 givenname: John surname: Bwanika fullname: Bwanika, John – sequence: 3 givenname: Mark surname: Okello fullname: Okello, Mark – sequence: 4 givenname: Davis surname: Rusoke fullname: Rusoke, Davis – sequence: 5 givenname: Faith surname: Nassiwa fullname: Nassiwa, Faith – sequence: 6 givenname: William surname: Lubega fullname: Lubega, William – sequence: 7 givenname: Davis surname: Musinguzi fullname: Musinguzi, Davis – sequence: 8 givenname: Alexander surname: Börve fullname: Börve, Alexander |
| BackLink | https://www.ncbi.nlm.nih.gov/pubmed/38223594$$D View this record in MEDLINE/PubMed |
| BookMark | eNptkUuLFDEUhYOMOA_duZYsXVhtkqpKJW5kGJxRGHBjr8OtPKojVcmYpBt66y83NQ9RcZXA_e45l3PO0UmIwSL0mpJN19LuPezy5sBazzaCP0NndGCyIT1pT-qfC9JwSfpTdJ7zd0IYp5K-QKetYKztZXeGfm6zDxOGVLzz2sOMfSh2nv1kg7Y4BmxsWqDEOU5HrGMwvvgYcsXwdoJg4AMGrCFb7GLCxh9syr4c13lJ4MOqbqAAzrbke2YBvfPB4tlCWscv0XMHc7avHt8LtL3-9O3qc3P79ebL1eVto1vRl8b0Y0uplSPlreyFcaI3xnBmaT9w5uQogDnOORAHo-WMmk7wkRLjBkFgcO0F-vige7cfF2u0DfXAWd0lv0A6qghe_T0JfqemeFCUDDUvIavC20eFFH_sbS5q8VnXtCDYuM-KSdqxeg3pK_rmT7PfLk_JV4A9ADrFnJN1SvsCa7ZrbHM1VWu9qtar7utVgteld_8sPen-F_8FI9urgg |
| CitedBy_id | crossref_primary_10_1007_s13671_024_00440_0 crossref_primary_10_3389_frai_2024_1394386 crossref_primary_10_1161_CIRCHEARTFAILURE_123_010879 crossref_primary_10_1177_11769351251371273 crossref_primary_10_1007_s13671_024_00436_w crossref_primary_10_1016_j_clindermatol_2024_06_016 crossref_primary_10_1016_S2468_1253_24_00053_0 crossref_primary_10_1016_j_sciaf_2025_e02783 crossref_primary_10_1186_s12929_024_01106_6 crossref_primary_10_1615_CritRevTherDrugCarrierSyst_2025053255 crossref_primary_10_62675_2965_2774_20250380 crossref_primary_10_1002_lio2_70152 crossref_primary_10_1016_j_compbiomed_2024_108294 crossref_primary_10_1007_s00403_025_04219_6 crossref_primary_10_1053_j_gastro_2024_03_004 crossref_primary_10_1007_s40615_024_02237_0 crossref_primary_10_1002_der2_70022 crossref_primary_10_1016_j_det_2025_05_003 crossref_primary_10_1016_j_hcl_2025_08_006 crossref_primary_10_1038_s41928_024_01213_0 crossref_primary_10_1053_j_gastro_2025_03_005 crossref_primary_10_2196_70903 crossref_primary_10_3390_healthcare11212906 crossref_primary_10_1016_j_imu_2025_101627 crossref_primary_10_1186_s12910_024_01008_0 crossref_primary_10_3390_diagnostics14212393 crossref_primary_10_3390_tropicalmed9010018 crossref_primary_10_3390_life14060652 crossref_primary_10_1007_s13671_025_00469_9 crossref_primary_10_1016_j_egyai_2025_100557 crossref_primary_10_1007_s12326_025_00721_w crossref_primary_10_1038_s41746_024_01399_9 |
| ContentType | Journal Article |
| Copyright | 2023 Kamulegeya L et al. 2023 Kamulegeya L et al. 2023 |
| Copyright_xml | – notice: 2023 Kamulegeya L et al. – notice: 2023 Kamulegeya L et al. 2023 |
| DBID | AAYXX CITATION CGR CUY CVF ECM EIF NPM 7X8 5PM |
| DOI | 10.4314/ahs.v23i2.86 |
| DatabaseName | CrossRef Medline MEDLINE MEDLINE (Ovid) MEDLINE MEDLINE PubMed MEDLINE - Academic PubMed Central (Full Participant titles) |
| DatabaseTitle | CrossRef MEDLINE Medline Complete MEDLINE with Full Text PubMed MEDLINE (Ovid) MEDLINE - Academic |
| DatabaseTitleList | MEDLINE - Academic MEDLINE CrossRef |
| Database_xml | – sequence: 1 dbid: NPM name: PubMed url: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed sourceTypes: Index Database – sequence: 2 dbid: 7X8 name: MEDLINE - Academic url: https://search.proquest.com/medline sourceTypes: Aggregation Database |
| DeliveryMethod | fulltext_linktorsrc |
| Discipline | Medicine |
| EISSN | 1729-0503 |
| EndPage | 63 |
| ExternalDocumentID | PMC10782289 38223594 10_4314_ahs_v23i2_86 |
| Genre | Journal Article |
| GeographicLocations | Uganda |
| GeographicLocations_xml | – name: Uganda |
| GroupedDBID | --- -OY 23M 2WC 4JU 53G 5GY 6J9 AAFWJ AAYXX ACGFO ADBBV AEGXH AENEX AIAGR ALMA_UNASSIGNED_HOLDINGS ANHLU AOIJS BAWUL CITATION CS3 DIK E3Z EBD EBS EJD EMOBN F5P GX1 HYE KWQ MK0 M~E OK1 OVT P2P RNS RPM SV3 TR2 C1A CGR CUY CVF ECM EIF M48 NPM OPA OVD RBI TEORI WOQ 7X8 5PM |
| ID | FETCH-LOGICAL-c385t-d5b311e9b163958df85ddd62e15762f9b8a2f666a0fabe621d486b10df780a7f3 |
| ISICitedReferencesCount | 34 |
| ISICitedReferencesURI | http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001039372900086&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| ISSN | 1680-6905 1729-0503 |
| IngestDate | Tue Nov 04 02:05:55 EST 2025 Thu Oct 02 07:04:57 EDT 2025 Wed Feb 19 02:08:44 EST 2025 Sat Nov 29 01:35:56 EST 2025 Tue Nov 18 22:38:26 EST 2025 |
| IsDoiOpenAccess | true |
| IsOpenAccess | true |
| IsPeerReviewed | true |
| IsScholarly | true |
| Issue | 2 |
| Keywords | Algorithms Fitzpatrick 6 skin type Telehealth Dermatology Artificial intelligence |
| Language | English |
| License | 2023 Kamulegeya L et al. Licensee African Health Sciences. This is an Open Access article distributed under the terms of the Creative commons Attribution License (https://creativecommons.org/licenses/BY/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. |
| LinkModel | OpenURL |
| MergedId | FETCHMERGED-LOGICAL-c385t-d5b311e9b163958df85ddd62e15762f9b8a2f666a0fabe621d486b10df780a7f3 |
| Notes | ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23 |
| OpenAccessLink | https://pubmed.ncbi.nlm.nih.gov/PMC10782289 |
| PMID | 38223594 |
| PQID | 2914257605 |
| PQPubID | 23479 |
| PageCount | -689 |
| ParticipantIDs | pubmedcentral_primary_oai_pubmedcentral_nih_gov_10782289 proquest_miscellaneous_2914257605 pubmed_primary_38223594 crossref_citationtrail_10_4314_ahs_v23i2_86 crossref_primary_10_4314_ahs_v23i2_86 |
| PublicationCentury | 2000 |
| PublicationDate | 2023-06-01 |
| PublicationDateYYYYMMDD | 2023-06-01 |
| PublicationDate_xml | – month: 06 year: 2023 text: 2023-06-01 day: 01 |
| PublicationDecade | 2020 |
| PublicationPlace | Uganda |
| PublicationPlace_xml | – name: Uganda – name: Kampala, Uganda |
| PublicationTitle | African health sciences |
| PublicationTitleAlternate | Afr Health Sci |
| PublicationYear | 2023 |
| Publisher | Makerere Medical School |
| Publisher_xml | – name: Makerere Medical School |
| SSID | ssj0026191 |
| Score | 2.462502 |
| Snippet | Background: In pursuit of applying universal non-biased Artificial Intelligence (AI) in healthcare, it is essential that data from different geographies are... In pursuit of applying universal non-biased Artificial Intelligence (AI) in healthcare, it is essential that data from different geographies are represented.... In pursuit of applying universal non-biased Artificial Intelligence (AI) in healthcare, it is essential that data from different geographies are... |
| SourceID | pubmedcentral proquest pubmed crossref |
| SourceType | Open Access Repository Aggregation Database Index Database Enrichment Source |
| StartPage | 753 |
| SubjectTerms | Artificial Intelligence Dermatology Female Humans Machine Learning Male Retrospective Studies Uganda |
| Title | Using artificial intelligence on dermatology conditions in Uganda: a case for diversity in training data sets for machine learning |
| URI | https://www.ncbi.nlm.nih.gov/pubmed/38223594 https://www.proquest.com/docview/2914257605 https://pubmed.ncbi.nlm.nih.gov/PMC10782289 |
| Volume | 23 |
| WOSCitedRecordID | wos001039372900086&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D |
| hasFullText | 1 |
| inHoldings | 1 |
| isFullTextHit | |
| isPrint | |
| journalDatabaseRights | – providerCode: PRVHPJ databaseName: ROAD: Directory of Open Access Scholarly Resources customDbUrl: eissn: 1729-0503 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0026191 issn: 1680-6905 databaseCode: M~E dateStart: 20030101 isFulltext: true titleUrlDefault: https://road.issn.org providerName: ISSN International Centre |
| link | http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwtV1LbxMxELbSghAXxJvwqIwEXKJtd-3drM0NoVQc0hShRMpt5d2121XbTWgelAsH_gv_k_FjHwlFKgcuq8i27NXOl_GMPfMNQm9YFioFfV4YK-6FVDBPUJV6UoC-pKEUqTA8s8N4NGLTKf_c6fyqcmHW53FZsqsrPv-vooY2ELZOnf0HcdeTQgP8BqHDE8QOzxsJ3gYB6A5HDlG0WTdnOvhVm6mWewm84byow8knJ_pcweY_Z7C9mRjEvI7c0CGRrqJET0eW9hZyaegcehcmJFNWNShO2iavrURUuozLnttymzskcbHSRxffjRU7nK2K5uD-myiLM7EdMXx8Vl0YtRONvqwWjjzUECe0TzMIbaKunAIGY9_THDV2f7qmzWltm6Xs0ElaKji25MPbWwMYSqHe904X-2tCC7J_HQP36Dg5nAyHyXgwHb-bf_V0cTJ9ie8qteygWySOuFaeRz8GtV8PPqjx66vXtPkVesGD9nKbls8f7sx2VG7LzBnfR_ecf4I_WFw9QB1ZPkR3jlwExiP008ALN_DCbXjhWYlb8MINvGAYtvB6jwXW4MIAHFyDS_dX4MIaXFiDy4xx4MIVuB6jyeFg_PGT5-p4eBll0dLLo5QGgeQp2P48YrliUZ7nfSIDcHaJ4ikTRIEbLXwlUtknQR6yfhr4uYqZL2JFn6DdclbKZwgT2DP6WRxywf2QSsEkD-IsUzCnAEcn6qJe9Y2TzJHc65c_T8DZ1RJJQCKJkUjC-l30th49t-Qufxn3uhJXAtpXX6mJUs5Wi4TwwLjsPqz81IqvnomC7U0jHnYR2xBsPUAzu2_2lMWpYXgPjOHO-PMbLPwC3W3-SC_R7vJyJV-h29l6WSwu99BOPGV7Bq-_ATyXzwk |
| linkProvider | ISSN International Centre |
| openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Using+artificial+intelligence+on+dermatology+conditions+in+Uganda%3A+a+case+for+diversity+in+training+data+sets+for+machine+learning&rft.jtitle=African+health+sciences&rft.au=Kamulegeya%2C+Louis&rft.au=Bwanika%2C+John&rft.au=Okello%2C+Mark&rft.au=Rusoke%2C+Davis&rft.date=2023-06-01&rft.issn=1729-0503&rft.eissn=1729-0503&rft.volume=23&rft.issue=2&rft.spage=753&rft_id=info:doi/10.4314%2Fahs.v23i2.86&rft.externalDBID=NO_FULL_TEXT |
| thumbnail_l | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1680-6905&client=summon |
| thumbnail_m | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1680-6905&client=summon |
| thumbnail_s | http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1680-6905&client=summon |