Purpose Blood oxygen saturation (SaO<InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="42452_2025_7129_Article_IEq1.gif" Format="GIF" Height="10" Rendition="HTML" Resolution="72" Type="Linedraw" Width="8" /> </InlineMediaObject> <EquationSource Format="TEX">$_2$</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mn>2</mn> <mrow /> </mmultiscripts> </math></EquationSource> </InlineEquation>), provided by blood gas analyzers, reflects the oxygen content in the blood. SpO<InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="42452_2025_7129_Article_IEq1.gif" Format="GIF" Height="10" Rendition="HTML" Resolution="72" Type="Linedraw" Width="8" /> </InlineMediaObject> <EquationSource Format="TEX">$_2$</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mn>2</mn> <mrow /> </mmultiscripts> </math></EquationSource> </InlineEquation>, provided by non-invasive methods, also indicates oxygen content. However, there is often a certain bias between SpO2 and SaO<InlineEquation ID="IEq3"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="42452_2025_7129_Article_IEq1.gif" Format="GIF" Height="10" Rendition="HTML" Resolution="72" Type="Linedraw" Width="8" /> </InlineMediaObject> <EquationSource Format="TEX">$_2$</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mn>2</mn> <mrow /> </mmultiscripts> </math></EquationSource> </InlineEquation> in practice. In the monitoring of oxygen levels in hospitalized patients, significant measurement errors due to differences in skin tone have been observed. Therefore, this study proposes a method to improve the accuracy of blood oxygen monitoring by fitting SpO<InlineEquation ID="IEq4"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="42452_2025_7129_Article_IEq1.gif" Format="GIF" Height="10" Rendition="HTML" Resolution="72" Type="Linedraw" Width="8" /> </InlineMediaObject> <EquationSource Format="TEX">$_2$</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mn>2</mn> <mrow /> </mmultiscripts> </math></EquationSource> </InlineEquation> measurement curves for populations with different levels of pigmentation. Methods We divided 71 volunteers into three groups based on the color of the skin on their hands: white, yellow and dark. We then used a fingertip pulse oximeter to control the oxygen content in the breathing mask in real-time, adjusting the blood oxygen levels of the volunteers. During the experiment, sampling points were recorded. The data processing consisted of two phases: First, we performed a group analysis of the obtained R values (the absorption ratio of red and infrared light) and SaO<InlineEquation ID="IEq5"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="42452_2025_7129_Article_IEq1.gif" Format="GIF" Height="10" Rendition="HTML" Resolution="72" Type="Linedraw" Width="8" /> </InlineMediaObject> <EquationSource Format="TEX">$_2$</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mn>2</mn> <mrow /> </mmultiscripts> </math></EquationSource> </InlineEquation> values. Second, we compared the differences in SpO<InlineEquation ID="IEq6"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="42452_2025_7129_Article_IEq1.gif" Format="GIF" Height="10" Rendition="HTML" Resolution="72" Type="Linedraw" Width="8" /> </InlineMediaObject> <EquationSource Format="TEX">$_2$</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mn>2</mn> <mrow /> </mmultiscripts> </math></EquationSource> </InlineEquation> measurements across different skin tones and compared them with the true SaO<InlineEquation ID="IEq7"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="42452_2025_7129_Article_IEq1.gif" Format="GIF" Height="10" Rendition="HTML" Resolution="72" Type="Linedraw" Width="8" /> </InlineMediaObject> <EquationSource Format="TEX">$_2$</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mn>2</mn> <mrow /> </mmultiscripts> </math></EquationSource> </InlineEquation> values. Results When analyzing the data from the 71 participants, we found that the skin-tone-specific calibration curves for pulse oximetry significantly reduced measurement errors and increased correlation. The correlation coefficients between measured and true values for the fitting curves were 0.9959 for the White group, 0.9977 for the Yellow group, and 0.9978 for the Dark group. For low oxygen saturation levels (SpO<InlineEquation ID="IEq8"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="42452_2025_7129_Article_IEq1.gif" Format="GIF" Height="10" Rendition="HTML" Resolution="72" Type="Linedraw" Width="8" /> </InlineMediaObject> <EquationSource Format="TEX">$_2$</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mn>2</mn> <mrow /> </mmultiscripts> </math></EquationSource> </InlineEquation> in the range of 70–80%), the mean absolute error for the White group decreased from 0.8947 to 0.5388. At moderate oxygen saturation levels (SpO<InlineEquation ID="IEq9"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="42452_2025_7129_Article_IEq1.gif" Format="GIF" Height="10" Rendition="HTML" Resolution="72" Type="Linedraw" Width="8" /> </InlineMediaObject> <EquationSource Format="TEX">$_2$</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mn>2</mn> <mrow /> </mmultiscripts> </math></EquationSource> </InlineEquation> in the range of 80–90%), the error for the White group decreased from 0.9010 to 0.4919, and for the Dark group, it decreased from 0.8379 to 0.2777. Conclusion This study successfully implemented skin-tone-specific calibration curves for pulse oximeters, accounting for variations in melanin content across skin tones. Results indicate that applying distinct calibration curves for African, Asian, and Caucasian groups significantly improved SpO2 measurement accuracy across all groups, with the most pronounced benefit observed in darker-skinned individuals. The use of this method can reduce the error due to skin color problems and improve the accuracy of blood oxygen measurement.

Calculation of arterial oxygen saturation by skin pigment adjustment: a research survey for people of three skin colors

期刊论文
出版模式：

开放获取
发表日期：

2025年8月26日

查看全文

Hui Pan,
Jilun Ye,
Zhiqiu Chen,
Xu Zhang

摘要

Purpose

Blood oxygen saturation (SaO $_2$ $_{2}$ ), provided by blood gas analyzers, reflects the oxygen content in the blood. SpO $_2$ $_{2}$ , provided by non-invasive methods, also indicates oxygen content. However, there is often a certain bias between SpO2 and SaO $_2$ $_{2}$ in practice. In the monitoring of oxygen levels in hospitalized patients, significant measurement errors due to differences in skin tone have been observed. Therefore, this study proposes a method to improve the accuracy of blood oxygen monitoring by fitting SpO $_2$ $_{2}$ measurement curves for populations with different levels of pigmentation.

Methods

We divided 71 volunteers into three groups based on the color of the skin on their hands: white, yellow and dark. We then used a fingertip pulse oximeter to control the oxygen content in the breathing mask in real-time, adjusting the blood oxygen levels of the volunteers. During the experiment, sampling points were recorded. The data processing consisted of two phases: First, we performed a group analysis of the obtained R values (the absorption ratio of red and infrared light) and SaO $_2$ $_{2}$ values. Second, we compared the differences in SpO $_2$ $_{2}$ measurements across different skin tones and compared them with the true SaO $_2$ $_{2}$ values.

Results

When analyzing the data from the 71 participants, we found that the skin-tone-specific calibration curves for pulse oximetry significantly reduced measurement errors and increased correlation. The correlation coefficients between measured and true values for the fitting curves were 0.9959 for the White group, 0.9977 for the Yellow group, and 0.9978 for the Dark group. For low oxygen saturation levels (SpO $_2$ $_{2}$ in the range of 70–80%), the mean absolute error for the White group decreased from 0.8947 to 0.5388. At moderate oxygen saturation levels (SpO $_2$ $_{2}$ in the range of 80–90%), the error for the White group decreased from 0.9010 to 0.4919, and for the Dark group, it decreased from 0.8379 to 0.2777.

Conclusion

This study successfully implemented skin-tone-specific calibration curves for pulse oximeters, accounting for variations in melanin content across skin tones. Results indicate that applying distinct calibration curves for African, Asian, and Caucasian groups significantly improved SpO2 measurement accuracy across all groups, with the most pronounced benefit observed in darker-skinned individuals. The use of this method can reduce the error due to skin color problems and improve the accuracy of blood oxygen measurement.